Professional Documents
Culture Documents
Stein Ringen - Citizens, Families, and Reform-Taylor & Francis (1997)
Stein Ringen - Citizens, Families, and Reform-Taylor & Francis (1997)
Families
& Reform
Citizens,
Families
& Reform
Stein Ringen
With a new introduction by the author
Originally published in 1997 by Oxford University Press
Notice:
Product or corporate names may be trademarks or registered trademarks,
and are used only for identification and
explanation without intent to
infringe.
Ringen, Stein.
Citizens, families, and reform / Stein Ringen ; with a new introduction by
the author.
p. cm.
Originally published: Oxford : Clarendon Press : New York : Oxford
University Press.
Includes bibliographical references and index.
ISBN 1-4128-0498-1
1. Social policy. 2. Family policy. 3. Child welfare. 4. Public welfare.
I. Title.
HN18.R47 2005
361.6'1—dc22
2005043675
PART I WELL-BEING
PART II REFORM
TRANSACTION EDITION
SCIENTIFIC DELIBERATION
The twentieth century made itself the century of social science. In the
process it also produced what was,
until close to its end, the authoritative
answer to the question of what the social sciences are for. That
answer was
contained in the idea of scientific management: once we understand human
behavior and
organization well enough we can manage human affairs with
scientific rationality.
For all the hopes that were invested in that idea, it finally proved itself a
monstrosity. It paid too little
attention to ordinary people. They are, it
turned out, not simple cogs in the machinery of society who will be
happy if
only managed well. Far from it; they are wilful individuals, often with
priorities different from
those of their fellows, and usually capable of
rationality on their own. Scientific management paid too much
attention to
extraordinary people, the custodians of scientific knowledge. They, the
experts – social engineers
in the work place, psychologists in the family,
economists in the economy, and the gentlemen of the central
committee in
governance – were the ones who were going to do the managing.
The twentieth century was also the century of democracy. But while
democracy prevailed, scientific management
collapsed. Indeed, scientific
management collapsed because democracy prevailed. The idea of
democracy
is that people should manage themselves because they know
best how they want their affairs to be arranged. The
idea of scientific
management is that they should be managed by others who know more than
they do. While still
believed in, that idea contributed to many ills and failed
experiments, such as taylorism, collective child
rearing, euthanasia, central
planning, autocracy, and dictatorship. As a result, scientific management
has
fallen into deserved disrepute.
That is all for the good, but it has left those of us who work in the social
sciences with a bit of a hangover.
If the idea that was, it must be confessed,
our idea was disgraced and if the social sciences were
capable of being
dangerous, what then?
In 1990, I was invited to take up the newly created chair in sociology and
social policy at Oxford University,
its first chair in sociology. I used my
inaugural lecture to make my contribution to the healing of our
collective
hangover. I am in no doubt that we are justified in defending our business as
a scientific
one.1 Nor about the importance of
this business to modern life.
But how should we make ourselves useful given that we want modern life
to be
lived democratically?
For my own answer to that question I took inspiration from an idea of
restraint. “Neither in politics
nor in science – no more than in love, family
or friendship – can we succeed for ourselves or be of use for
others if we
always insist on holding our ground.” Scientific management was social
science unrestrained, the
social scientist as the arbiter of truth and wisdom.
We social scientists have, I think, good reasons to be
assertive about the
contribution we can make towards order in society. My reason for saying
that is not miles
away from scientific management: there are some things
we should know better than others. But there is also
much we cannot know
better than others and therefore we should offer our services in a spirit of
restrained
assertiveness.2
My argument was essentially a democratic one. Since then, we have seen
some important advances in democratic
theory under the label of
“deliberative democracy.”3 That enables me now to give the program I was
suggesting then a proper name. It is
diametrically opposed to scientific
management, yet not miles apart. I call it scientific
deliberation.
The theory of deliberative democracy says that rationality in collective
decision-making and governance depends
on more than the aggregation of
preferences in elections and similar arrangements, it rests as well on the
honing of those preferences through continuous processes of deliberation.
What we want to do in those processes
is to cut through the fog, identify the
matters that really deserve disagreement and then try to work our way
through to some kind of shared platform or reasonable compromise where
we recognize the inevitability that
there is no single compromise that is the
best one for all concerned. Deliberative democracy is an
eminently
idealistic model of citizens of good will working earnestly with each other
to make themselves
enlightened and work their way through their inevitable
differences in a civilized manner. It is the kind of
idealism that appeals to
someone like myself who believes that reason comes from restraint.
Productive deliberation depends crucially on information. Obviously, it
depends on much more—effort
and honest good will—but information is
essential. Although no sharp dividing line can be drawn between factual
and moral questions, we are in a much better position to concentrate debate
on the predominantly moral ones,
which are the questions that deserve
debate, if we know as much as possible about the predominantly factual
ones. Information is obviously no less necessary in actual decision-making
than in deliberations before and
after decision-making, but the theory of
deliberative democracy helps us to understand better the significance
of
informed democracy.
Informed democracy, in turn, depends on conditions of its own. A first
one is that citizens have access to the
best quality of relevant information.
For that they should turn to the scientists; science is not
omni-
knowledgeable nor infallible but the scientific method is the best we have.
The scientists, therefore,
have the capacity to be of service by contributing
to clearing away the fog of ignorance. A second condition is
that
information be free and not under the control of parties or interests who
might use it manipulatively. The
scientists, therefore, have a duty, in this
regard, to inform democracy objectively.
Information, I said in that lecture, is “the common ground on which
science and politics can meet without
scientists playing politicians or
politicians playing scientists; information is what politics needs and
science
can provide.” That led me to a three-item program for the contribution of
social research to scientific
deliberation: First, as social scientists we should
accept responsibility for informing the democratic
process. Second, we
should not only accept that responsibility but also claim this as our
authority.
Third, although we should claim the domain of information as our
own, we should not, as scientists,
transgress into the domain of decisions.
The lecture that contains this program is reproduced as the last chapter in
this book. The rest of the book is
dedicated to doing what I said there
should be done. I take on two issues on which I think we social scientists
have not done our part of the job towards scientific deliberation. I could
have called the book “The
Truth About Families and the Truth About
Inequality.”
When I say “we social scientists,” I’m afraid I mean “we sociologists.”
Although the family is the sociological
institution par excellence, family
sociology has been wholly unsuccessful in ridding political
deliberation of
one of its great myths. That is the belief that modern families – or the
family as is
often the terminology – are typically small nuclear families,
that the nuclear family has arisen from the
demise of the large extended
family, and that while extended families were multi-functional, nuclear
families
are “reduced” to specializing in catering to the emotional needs of
family members. In my native language,
Norwegian, the concept is
funksjoristϕmming, the family bereft of functions.4
I am sure it would be unfair to accuse family sociology of having created
this myth, but it is regrettable that
it has not been able to kill it. Still today,
it is impossible to listen to any debate about family policy
without the
mythology of “the nuclear family” being offered up as if it were obvious
truth.5
The matter is important. If the nuclear family story is believed, the
recommendation in social policy would be
to build up alternative
institutions that could take on the functions families can no longer perform.
If that
were to happen, social policies might contribute to the family demise
that is thought to have already happened
and we would be in a spiral of self-
fulfilling prophesies.
Since I had become distrustful of family sociology – there is altogether
too much “discourse” and too little
hard-nosed empiricism – I turned
instead to the tools of economic analysis. That allowed me to ask and
answer
the question of what families do. And what I found was that modern
families in modern societies in
fact do astonishingly much, that much of
what they do is elementary economic production, and that therefore
they are
not sites of “only” social and emotional life.
While family is a sociological favorite, inequality is another one. One
myth about inequality, very widely
believed in European sociology today, is
that changes in social structure which we often associate with
progress in
our part of the world, such as economic growth or the expansion
of access to higher education, contrary to expectations tend not to filter
down to less inequality
between persons or groups in well-being or
opportunity.
If that story were true, its influence in public policy could be drastic and
unfortunate. We would, for
example, have to take it that a policy of opening
higher education up to more young people would probably make
little or no
contribution towards a more open society. Exactly that scenario is presently
unfolding in Britain.
The government has announced that it wants to extend
university-level education enough to admit 50 per cent of
graduates from
secondary school (from short of 40 per cent today). That policy is meeting
massive resistance (a
resistance which is really only understandable in light
of a deep skepticism to education in British culture).
The resistance comes,
predictably, from taxpayers and some of those who see themselves as their
representatives, but also, less predictably, from a mixed lobby of
commentators who sometimes argue that the
new graduates would not be
able to benefit from university training and sometimes that the
consequences would
be a levelling down of standards.
The myth about inequality that I have referred to plays a minor part in
this drama. If it cannot be expected
that more places in university education
will contribute to less inequality, then from that point of view there
is no
reason to go down that road.
I have for years been concerned with issues of public policy, inequality,
and welfare. In that career, during
the last twenty years or so, I’ve done
battle with a parade of propositions that by and large see inequalities
to be
so entrenched in economic and social organization that they are beyond the
reach of reformist public
policy. As a result of this feisty warfare, I believe
that I have been able to pick off those propositions one
by one.
First proposition: The distribution of income in industrial societies tends
towards stability and is not
sensitive to (weak) policy interventions, at least
in the long run. My reply (in The Possibility of
Politics): Advances in
comparative research on income since about 1970 have shown that income
distributions fluctuate strongly within nations and differ strongly between
nations, and that such fluctuations
and differences are associated with
changes and differences in government policies.6
Second proposition: Although the distribution of income is not stable, the
incidence of poverty is, the
implication being that reformist policies are not
sufficiently powerful to rectify injustices at the bottom of
the distribution.
My reply (in “Direct and Indirect Measures of Poverty”): This proposition
has been shown to
rest on weak definitions of poverty that make empirical
results non-interpretable, not sufficiently robust for
alternative methods of
measurement, and unable to stand up to stricter and more appropriate
criteria of
definition and measurement.
Third proposition: Although the distribution of income and the incidence
of poverty are not insensitive to
reformist policies, class inequalities of
opportunity are, the implication being that deep, hard-core
inequalities
persist unabated. My reply (in chapter 4 below): Support for this last stand
of the
anti-reformists has been sought in elaborately estimated statistics that
propose to show long-term stabilities
in class inequality. These statistics
have, however, proved no less fragile than those on stable income
distributions and poverty rates. Stability results rest on excessively
relativized methods of measurement that
are biased in favor of the preferred
hypothesis, the results are not robust enough for tests by alternative
methods, and are disconfirmed by methods of unbiased measurement.
The first two propositions have been successfully despatched and those
matters can be considered settled. The
third one, however, is still believed
to be true and therefore deserves another round of attention.
It turns out that economic analysis is ideally suited to uncover what family
sociology has found it so
difficult to describe: what families do.7 Of course
that analysis can only tell us what goes on economically within
families, but
that is just what we need to know for our purpose. The theory of
funksjonstϕmming will
have it that less is done in families precisely
because they no longer perform economic functions.
What made it possible to analyze within-family economies with a detail
and authority that was innovative at the
time, and perhaps still is, was the
availability of two British data sets that Stephen Jenkins, Jay Gershuny,
Brendan Halpin, and I were able to combine through a process of synthetic
merging, one on family income and one
on family time-use. I was able to
use those merged data to measure the volume of within-family production
and
to add the value of that production to the value of goods families could
buy in the market for their cash
income. I was also able to introduce
assumptions about economic co-operation in families and to impute values
to the gains that are made through economies of scale when total goods
available to families are metamorphosed
into consumption value for family
members.
All of this is, of course, quite technical, as can be seen in chapters 1 to 3,
but the long and the short of it
was a conclusion, based on cautious and
conservative assumptions and understating rather than overstating,
that, in
these modern families in this modern society, within-family economies of
production and co-operation
add as much value to final consumption as the
value of goods and services consumers can buy in the market for
the
income available to them. If the national economy is what produces a
population’s capacity for consumption,
half of the nation’s economy is in
the nation’s families.
That conclusion, which I think is absolutely robust, was and is, in my
opinion an astonishing one. If we think
modern families are bereft of
functions, we are simply wrong. If we think modern families are no longer
economic institutions, we are simply wrong. More generally, if we think
that our well-being, even defined in
narrow economic terms, is generated in
formal settings “out there” in society, such as in markets, and then
taken
home to the informal setting of the family for enjoyment, we are also
wrong. What we take home from
markets are raw materials that are
processed further in the family before we sit down and consume the
combined
product of market and family economies. That processing in the
family is no-nonsense economic enterprise: it is
adding value to raw
materials by applying labor force and capital to them and it is sharing and
distributing
the produce.
According to my program of scientific deliberation, my job is to establish
facts such as these but not to go on
and draw conclusions for what should
be done in social policy. That’s for deliberation. What the facts say is
that
modern families are not bereft of functions and that they are what families
have always been: vibrant and
essential institutions in our quest for well-
being. If, however, I were to disregard my own rules of scientific
deliberation and transgress into the domain of decisions, which I of course
do, I would say for social policy:
invest in families! The truth about
families is that they are the most productive institutions known to
mankind.
NOTES
1 But in my own university that was only formally accepted when, in a
reorganization in 2000, what had been the faculty of social studies became
the division of social
sciences.
2 This idea of restraint I had applied previously in an analysis of good
families, see,
e.g., Mason et al. 2003.
5 This is the true story: “The delusion about marriageable age is,
perhaps, the
most conspicuous of all the errors we seem to want to make
about everyday life in the old world, but is not the
only one. More far
reaching in its consequences for the view we take of ourselves, and the
plans we make for our
society and its welfare, is the supposition that the
family group in the pre-industrial world was large. It is
erroneously
believed that it was large because it contained whole groups of kinsfolk
living together… In any
case, and this seems to be the most deep-seated and
important generalisation of all, in the familial, patriarchal
world the family
was seen as the source of welfare. Sickness, unemployment, bereavement
were all the
responsibility of the family, and so to a large extent was
education. To fulfil all these functions the family in
the old world would
have to be large. Something like this seems to be the general impression of
family life among
our ancestors. But their families were not large, at least
the average family was relatively small. In
fact the evidence we now have
suggests that household size was remarkably constant in England at all
times from
the late sixteenth until the early twentieth century.” (Laslett
1971, p. 93).
6 The Possibility of Politics was first published in 1987 and is to be
republished with a new introduction in 2006 by Transaction Publishers.
7 Another case, regrettably, of economics, with its commitment to
working
scientifically, crowding out sociology, with its sometimes
orientation to “discourse,” even on sociology’s home
turf.
8 Goldthorpe et al. 1987, p. 328.
9 Erikson and Goldthorpe 1992, Marshall et al. 1997.
10 Goldthorpe 2000. p. 163, 257.
11See, e.g., Shavit and Blossfeld 1993.
12 That seminar was in Nuffield College in 1992. I chaired it. Professor
Pahl, a
distinguished colleague, was treated in a shoddy manner. I have
blamed myself for allowing a seminar I chaired to
disintegrate the way it
did. The experience helped me to decide that I would try to get to the
bottom of what my
intuition told me had to be an unresolved confusion in
this branch of class analysis and that the only way to do
that would be to
penetrate the complexity of its methodological basis.
13 Hellevik 1997, later followed up in Hellevik 2002 and in further as
yet not
published work.
14 On the methodology, see Erikson and Goldthorpe, 1992, ch. 2.
15 The concept of ‘margin-insensitivity’ turns out to be rather mysterious
both
in origin and meaning, cf. Hellevik 2002.
16 See Marshall and Swift 1999, replies by Hellevik 2000 and Ringen
2000, and
counter-reply by Marshall and Swift 2000.
17 Ringen 2000, p. 84.
18 To see just how murky and complicated it is, consult Hellevik 2002.
19 In the literature, this “something” is called by a range of more or less
merits being told. The research program to which Goldthorpe attributes “the
main achievement to date of class
analysis” has in fact come under strong
and sustained criticism. Adam Swift in his summary article very properly
takes on that criticism and ends up salvaging the research program by
underlining its limitations and narrowness.
In On Sociology, Goldthorpe
deals with that same criticism by simply not mentioning it. Swift refers and
responds to criticism by Pahl, Sorensen, Holton, Crompton, Saunders,
Hellevik, and Ringen. None of the relevant
works are mentioned by
Goldthorpe. The introduction (to what is a collection of essays) does not
take the
opportunity to respond to this criticism and continues to generalize
from findings that according to Swift are
“narrower or more carefully
specified than some had realised” to conclusions about class inequality in
all
generality: “Despite the expansion of educational provision and
increases in general levels of attainment that
are apparent in all modern
societies, class differentials have in most cases remained rather little altered
over
many decades” (p. 23). Pahl is accorded the same respect as at his
seminar. He gets one mention, not his critical
article but a short “reply to
Goldthorpe and Marshall.”
REFERENCES
I started the studies that have trickled into this book when I joined the
University of Oxford in 1990. I have
been able to discuss the work in its
various stages with my students, and their sharp and critical, and often
unexpected, responses and suggestions have been helpful more than they
may know.
I wish to express my gratitude to the many colleagues who have offered
advice, and sometimes material, often
taking time to read and comment on
drafts, including Tony Atkinson, Raymond Boudon, Jerry Cohen, Roger
Crisp,
Partha Dasgupta, Maurizio Ferrera, Diego Gambetta, Anne Gauthier,
Jay Gershuny, Anne-Marie Guillemard, Erik
Jϕrgen Hansen, Anthony
Heath, Michael Hechter, Duncan Ironmonger, Stephen Jenkins, Nanak
Kakwani, Jon
Eivnd Kolberg, Robert Lane, Michael Lockwood, Claus Offe,
Nigel O’Leary, Ivo Možný, Martin Potůček, Natalie
Rogoff, Tor Rϕdseth,
Peter Saunders, Hannu Uusitalo, Mariken Vaa, Jiří Večerník, Claire
Wallace, Kari
Wærness, and my brothers Anders and Knut. The work on
class inequality behind Chapter 4 was undertaken jointly with Ottar
Hellevik, of the
University of Oslo. He found the ingeniously simple
devices for getting to the core of the terribly difficult
matter of analysing
inequality from mobility tables. The chapter here is one of several products
of that
cooperation. We owe special thanks to Anthony Heath, who, with
exceptional generosity, helped us with access to
the reanalysis of his own
data on class and education. Brendan Halpin was my research assistant
from 1993 to 1995
and undertook the empirical work for Chapters 2 and
3,
including solving many of the methodological problems
in the Appendices
to Chapter 3. Many more than I can
mention have contributed ideas and
suggestions, not least in a range of seminars I have had opportunity to give
far and wide during the process. Susan Dyson, Yvonne Painting, and Susan
Walters have been my outstanding
assistants in all matters of University
work and made it possible for me to do some research and writing and
helped me in countless ways in those efforts. Early ideas for the works
contained in Part I were presented in several seminars at the Social Policy
Research Centre of the University of New South Wales in Sydney while I
was a Visiting Fellow there in 1991. The
final editing of the manuscript was
undertaken in the beginning of 1996 while I enjoyed some peaceful weeks
as
professeur invité at the University of Paris, Panthéon-Sorbonne. This is
an opinionated book—it even
offers the opinion that social scientists should
not be opinionated—and it is therefore in place to state the
obvious that
these friends and colleagues carry no responsibility for opinions, nor for
such defects as there may
be in analyses.
Chapter 1 is a somewhat revised version of a paper
published previously
in Acta Sociologica (1995, no. 1). Chapter 5 contains material also used in a
paper in the Czech Sociological Review (1996, no.
1). Chapter 6 is, with a
few cosmetic changes, my
inaugural lecture at the University of Oxford
(delivered 27 October 1992, published separately by OUP in 1993). I
am
grateful to editors and publishers for permission to reprint and use
previously published material.
The work reported in Part I was supported by a grant from the
Economic
and Social Research Council (grant R000234427).
S.R.
Oxford
29 February 1996
PREFACE:
REASON AND RESTRAINT
Suddenly I had the absurd feeling that at last I understood Lenka Silver’s
enigmatic essence: in spite of her
daring bikini, in spite of her painted
fingernails, she was actually living at the turn of the century, at a
time
before all values had become relative.
Josef Škvorecký, Miss Silver’s Past
and Sverdrup
1995.
3 Preston 1984.
4 Palmer et al. 1988.
5 House of Representatives 1993; in 1991, almost one in four American
chil-
dren—22 per cent—lived in poverty thus defined.
6 DSS 1994.
7 INSEE 1996.
8 Smeeding et al. 1990, although Britain is an exception to this rule.
9 I see no grave problems with extending the vote to children, except for
this:
in some societies, children make up much more than about a quarter of
the population. Even I would be reluctant
to accept a below-18 majority in
the electorate. This, of course, is not an argument against resolving the
democratic deficit arising from the exclusion of children, but it might be an
argument for further compromise in
how to arrange the political
representation of children.
10 Boudon 1995: 46–7.
11 Goldthorpe et al 1987: 329.
12 Cf e.g. Dasgupta 1993.
13 While waiting for proofs. I am reading Isaiah Berlin’s The Crooked
Timber
of Humanity and there find this (p. 18): ‘… the search for perfection
does seem to me a recipe for bloodshed
...’
PART I
Well-Being
INTRODUCTION: HARD FACTS
ABOUT SOFT VALUES
PERSONS
People have different preferences for arranging their lives. Prefer- ences
should be respected by researchers who
want to compare how well people
live and by politicians who want to plan for a good society. The researcher
is in
no position to decide for people how they ought to live their lives and
should not offer criteria for government
policy with a built-in dictatorial
logic.
Approaches to well-being can be welfarist or non-welfarist. In welfarist
approaches, the individual’s experience
of how well he or she lives is taken
as the criterion of well-being. This is the approach of mainstream
theoretical welfare economics and of psychological quality-of-life research.
In non-welfarist approaches,
well-being is established from the ‘objective’
circumstances within which people live rather than from their
‘subjective’
utility, satisfaction, or happiness. Most empirical research on inequality in
economics and
sociology is in this sense non-welfarist.
It is often implied that welfarist approaches respect preferences while
non-welfarist approaches assume
preference agreement. The cost of
respecting preferences is taken to be the impossibility of valid interpersonal
welfare comparison, while the benefit of ignoring preferences is the
possibility of getting on with empirical
measurement.
This gives rise to two tensions. On the theoretical level, a tension is seen
between welfarist approaches, which
are taken to have the virtue of being
non-dictatorial, and non-welfarist approaches, which are considered to lack
this virtue. The second tension is between theory and measurement; while
the non-dictatorial virtue can be
preserved in theory, it is sacrificed in
measurement, either because experienced well-being (utility) is assumed
not to be directly observable or because data are available on people’s cir-
cumstances only.
A possible solution is to stick to welfarist approachs in theory, but give
up the ambition of developing
empirical measures which are at least close
to capturing underlying well-being. In this
case, a study of income
inequality, for example, would be simply a study of income inequality and
not of the
distribution of well-being. This solution is, however, not only
theoretically shallow but has also been overtaken
by events in empirical
research. For example, when household income is re-estimated to
equivalent income, which
is now customary, we are no longer studying
physical income but well-being as measured by a highly theoretical,
adjusted income measure (see Chapter 2). A second
solution is to develop
methods for measuring experienced well-being directly. This is possible, as
demonstrated
in quality-of-life research,1 but I have
reservations about such
measures, for reasons which will be explained below. I propose a third
solution in which
the welfarist/non-welfarist tensions are resolved without
resort to measures of experienced well-being.
I argue along four lines. I start from a notion of well-being as freedom,
which I hold to be both
ethically sound and methodologi- cally simple in
the sense that only modest assumptions are needed about what
people seek
in order to live well. I introduce the con- cept of arenas between resources
and outcomes,
and offer a speci- fication of the circumstances of choice as a
function of personal resources and arena options.
I draw a distinction
between prefer- ences and expectations and with the help of that
distinction
show that welfarist approaches do not have a special quality above non-
welfarist approaches of
respecting preferences. I offer a typology of
approaches to the measurement of well-being and with the
help of this
typology show that non-welfarist measurement can be or- ganized so as to
be neutral with regard to
preferences.
I propose to demonstrate that non-welfarist/non-dictatorial approaches to
(the measurement of) well-being are
possible. The most powerful non-
welfarist approaches in the literature are in terms of goods (Rawls and
others),
resources (Dworkin and others), and more recently capabilities
(Sen). All have deficiencies. Goods approaches pay
too little attention to
preferences. Re- sources approaches pay too little attention to structural
constraints on
the individual. The capabilities approach has the potential of
overcoming these problems, but is shot through
with ambiguities which
need to be sorted out.
A TYPOLOGY OF APPROACHES
Direct approaches
Typology
Indirect Direct
Narrow (1) The income approach (4) The expenditure approach
(2) The resource approach (5) The consumption approach
Broad (3) The capabilities (6) The way-of-life approach
approach
PREFERENCE NEUTRALITY
The argument for welfarism is that people should have the au- tonomy to
decide what is good for themselves, and
that personal satisfaction is
therefore the best criterion of well-being. The criti- cism of non-welfarism
is
that this autonomy is violated and that a dictatorial element is introduced
by implicitly imposing a common
utility function on all persons. Neither
point is well taken.
Every person has a ‘filter’ in his or her mind through which the
‘objective’ circumstances he or she lives in
are translated into ‘sub- jective’
utility, satisfaction, or happiness. There are two elements in this filter,
on
the one hand expectations with regard to the level of consumption, and on
the other hand
preferences with regard to the composition of consumption
(be it consumption in the narrow
‘economic’ understanding or taken
broadly as way of life). If two persons have different expectations with
regard to the level of consumption, they will need different levels of income
to achieve the same level of
satisfaction. People may, however, want to live
very different lives if they have different preferences for the
com- position
of consumption without, therefore, requiring different levels of income.
Utility, then, (or
satisfaction or happiness) is a product of circumstances,
expectations, and preferences; not of circumstances
and preferences only.
The two components I here call preferences and expectations are often
treated as a single category, then called
tastes. If the issue of autonomy is
seen in terms of tastes, one encounters the problem of expensive tastes,
i.e.
that some people insist they can only be happy if they drink champagne and
drive Rolls-Royces (and, for
that matter, a problem of inexpensive tastes,
i.e. that some people will be happy even if they live in
squalor). True
welfarists will have to give in to expensive tastes. Dworkin has
argued that
this makes nonsense of welfarist approaches to well-being and that tastes
should be disregarded in
favour of a pure resource approach.14 Cohen
has
argued that expensive tastes which people have through no choice of their
own should be respected and
therefore that some semi-welfarist approach is
needed.15
I suggest that this broad category of tastes is not very meaningful. It is a
totally different matter to want a
higher level of consumption compared to
wanting a different composition of consumption. Nor do I think the
notion
of expensive tastes as chosen or not chosen has much to it (and what about
inexpensive tastes that are
not chosen?). Where tastes come from, we do not
know, but tastes are hardly things we simply choose.
The broad category of tastes needs to be broken down into preferences
and expectations. While preferences
should indeed be treated as tastes,
expectations should be treated as demands or interests.
Within a liberal agenda, each person should have the right to choose a
composition of consumption (or a way of
life) according to his or her own
preferences. This must, however, be done within some budget limitation (or
more generally, with due concern for the rights of others to choose their
way of life). While preferences
relate to the composition of consumption
within a budget, expecta- tions relate to the size of the budget
itself. The
reason the choice of composition should be free is that my choices on how I
spend a given budget do
not (in principle) affect your budget. The reason
the choice of budget cannot be free is that the more I take
the less you can
have. The distinction between preferences and expec- tations thereby allows
for a liberal
theory in which preference autonomy is respected within the
limitation of concern for the preferences of
others, and reveals true
welfarism to be not liberal but libertarian, implying the acceptance of free
choice
without concern for the consequences for others.
Since personal satisfaction is, in part, a product of expectations, we
cannot accept satisfaction as the
criterion of well-being without also
accepting the individual’s right to determine what level of consumption
(or
budget) he should have. If we do not accept the latter, as I think we should
not, neither should we accept
the utility approach to well-being. Unless you
are willing to be blackmailed by the
rich kid who demands himself to be
rich because he says he cannot be happy without the style of life he has
been accus- tomed to, or to ignore that the inability of the deprived to
articulate their deprivation is often
part of their deprivation, you abandon
utility as the test of well-being. The rejection of utility does not
imply a
rejection of the individual’s autonomy to follow his or her own preferences.
Utility is rejected
because of the problems caused by expectations, and not
because of any difficulty with preferences.
Having now dismissed welfarism as the approach associated with
autonomy of preferences and having demonstrated
that this does not imply
the introduction of a dictatorial element, my next question is whether non-
welfarist
measurement is possible without violating preference autonomy.
The alternative to measur- ing well-being as the
product of circumstances
and preferences (which, anyway, cannot be done since utility is
‘contaminated’ by
expectations) while still respecting personal preferences,
is to use information on circumstances alone but to
organize this informa-
tion so that no priority is given to certain preferences above others. The
ability of a
measure not to violate personal autonomy with regard to the
composition of consumption, or more generally, with
regard to how people
choose to arrange their lives, I call preference neutrality.
With indirect measurement, well-being is measured by the choices people
can make. In this, no
assumption is imposed about what choices people
should make or about some choices being superior to others. In
principle,
therefore, these approaches are preference-neutral.
With direct measurement, well-being is measured by the out- comes of
the choices people have made.
This requires stronger assumptions about
preferences. In order to evaluate how well people live on the basis of
the
choices they have made, we need assumptions about how they want to live,
or should want to live, against
which how they in fact live can be compared.
For example, if we measure well-being by consumption, we give
priority to
con- sumption preferences over the preferences of misers for saving and
givers for charity. Direct
approaches, therefore, are in principle not
preference-neutral.
Although the distinction between direct and indirect measure- ment is
logical when starting from the concept of
choice, only indirect measurement
approaches are ultimately compatible with a liberal
choice perspective on
well-being.
Some qualifications are needed. In the income approach, it is implicitly
assumed that people agree about the
importance of in- come and the relative
importance of goods which can be bought relative to goods which are not
bought. There are, therefore, limitations in the preference neutrality of this
approach.
On the side of indirect measurement, preference neutrality is improved
when the approach is broadened. In the
capabilities ap- proach, there is in
principle perfect preference neutrality since the full range of choice is
described in terms of both personal resources and structural options, i.e.
actual, accessible opportunities.
In the expenditure and consumption approaches, consumption
preferences are given priority over non-consumption
preferences, but these
approaches are still neutral with regard to different pref- erences within the
range of
preferences considered. If, for example, total consumption is the
same, people who spend their money on the
town are considered no better
or worse off than people who spend their money on home and family.
On the side of direct measurement, preference neutrality is re- duced
when the approach is broadened. The
way-of-life approach can be
interpreted as a specification of the arguments in the welfare function which
is
assumed shared by all persons. However, even this is not to assume full
agreement on preferences. Agreement is
assumed on the arguments in the
welfare function but not necessar- ily on any specific combination of
arguments. If well-being is meas- ured by considering, consumption,
wealth, and social integration, there may
be a range of different but
equitable combinations. It is perhaps not a very radical assumption that
there is
in any culture a fairly broad agreement on what basic factors are
important for well-being, but there will not
be total agreement, and in multi-
cultural societies there may be much disagreement. The prefer- ences of
those
who are not part of mainstream agreement will be disregarded, and
this is perhaps what respect for personal
preferences is all about.
DISCUSSION
Interpersonal comparison
Redistribution
The conceptual pair of resources and arenas has very considerable power. It
offers an
operational conceptualization of the circum- stances of choice in
terms of real, available, and accessible
oppor- tunities, and thereby rescues
the ‘freedom of choice’ slogan from the narrow understanding it has
taken
on in much of the literature as little more than the absence of political
interventions in markets.
This is a conceptualization with broad usefulness,
equally applic- able to poor and rich societies and to the
situation of poor
and rich people, and for capturing the life chances of women and men, the
young and the
old, and the handicapped and the strong. It shows up the
limitation in pure resourse approaches—that
structural con- straints are
ignored—and also how that limitation can be over- come. It gives a
framework for
monitoring real opportunities without, as in goods
approaches, having to rely on information on outcomes
after people have
made their choices on how to organize their lives.
Capabilities
Levekårs-
undersøkelsen (1976).
10 e.g. Sen 1985, 1992.
11 Cf. e.g. Goldschmidt-Clermont 1982, 1993.
12 Cf. e.g. Gillespie 1965.
13 Cf. Becker 1965.
14 Dworkin 1981a, b.
15 Cohen 1989.
16 Cf. e.g. Griffin 1986.
17 Cf. e.g. Erikson and Uusitalo 1987.
18 Williams 1987; Cohen 1989.
19 Sen 1992.
20 As also Arneson (1989) has argued.
2
FAMILIES
There are two units of analysis: the family and the individual. Analyses of
goods and well-being move back
and forth between these two types of units.
Goods are the raw materials of well-being. Goods are not con- sumed
directly by persons, but are used by and
distributed in families. Hence, in the
analysis of goods, the family is the first-order unit of analysis. (We
may in
the next round turn to the individual as the unit, for example if we want to
know the volume of goods
per person.)
Well-being is observed in the status of family members, after family
economies. In the analysis of
well-being, the individual is the first-order unit
of analysis. There are three reasons. First, empiric-
ally, the end outcome of
family economies are not known until it is known how these outcomes are
absorbed by
family members. Sec- ond, philosophically, the phenomenon of
well-being resides with individuals and with
individuals only. Groups, from
families and upwards to populations, do not have well-being in any other
mean-
ing than as some aggregate of the well-beings of the members of the
group. Third, normatively, only with the
individual as the unit of analysis (or
by weighting each family by the number of members, which is the same
thing) can each person in the population be given the same weight (count
equally) in determining the social
condi- tion. This is the democratic
approach to the measurement of well- being, akin to the one-person,
one-
vote principle in politics.
The processes of family economies between market goods and well-being
are illustrated in Figure 2.1. In single-person house- holds, production and
needs may intervene between market goods and well-being. In multi-person
households, also co-operation and
distribution apply. Generally, family
economies add value, al- though theoretically value might be lost if
families
waste rather than produce goods or fight rather than co-operate. (Various
possible feed-back processes are not considered, for example from co-
operation or distribution to
production.)
Implications
EQUIVALENT INCOME
EMPIRICAL ILLUSTRATIONS
A range of estimates for Britain in 1976 and 1986 are given in Table 2.1,
using Family Expenditure Survey data, adjusted for over-time
comparability.6 Equivalization is by the OECD scale. All estimates
are based
on disposable income (1986 £s). Household production is disregarded.
Within-household distributions are assumed equit- able. Precise definitions
are given in the table.
The intention is to illustrate (1) value added, (2) the difference between
economic growth and trends in
well-being, and (3) the difference between
valid and invalid measures of inequality of well-being.
Value added
Economic growth
Inequality
Since the valid alternative equivalence scales are limited to those with the
same structure but either
flatter or steeper profiles, the consequences of
applying alternative relevant equivalence scales are clear
enough. The use of
a flatter scale would in all cases reduce the difference between the results of
preferred
and alternative measures—less value added, higher rates of change
in average well-being, higher levels of
inequity, and lower rates of change in
inequity. The use of a steeper scale would have the opposite results.
CONCLUSIONS
Absolute interpretation
Family economies
Once the ‘black box’ of the family is opened, the economic life within
families turns out to be remarkably
complex. There is pro- duction and
distribution, and between goods and well-being value is added through the
joint use of goods, as mediated through needs. In economic parlance, the
term ‘production’ is often used for
the entire process of translating goods
into well-being. In families, however,
there are at least three distinct
processes which need to be identified separately: a process of creating
goods
(which I call production), and a process making goods public (which I call
co- operation), and a
process of distributing services from goods.
In terms of consumption well-being to persons, as much value may be
generated in the average family as the
value that flows to the family through
market activities.
Inequality in Britain
Economic growth
Children
family
theme of the book, except in the empirical illustrations in this chapter,
in which I use household data. Much of
what is said here, would apply
equally to families and households, although there might be some
differences in
the patterns of economic co-operation between families and
non-family households.
2 Buhmann et al. 1988; Coulter et al. 1992.
3 Cf. e.g. Deaton and Muellbauer 1980; Nelson 1992.
4 OECD 1982.
5 Atkinson and Michlewright 1992.
6 Coulter et al 1994.
3
CHILDREN
Children are vulnerable citizens; they are dependent on the good- will of
others and have little power of their own. Children are also important
citizens; they represent the future of their society. How a society treats its
children has much to say about its charitable quality and its rational
organization.
This is a study in the well-being of British children. It covers the period
from 1976 to 1986, which were years of rapidly increasing income
inequality. According to the most authoritative study of changes in the
distribution of income in Britain, the year 1977 was a turning-point
whereafter inequality increased steadily at least up to the beginning of the
1990s.1 The substantive questions in the study are about how children and
families with children fared during a period of overall increasing inequality.
It is also a study in measurement. The income approach to the
measurement of well-being is an important and much used approach which
has a number of strengths, in spite of well-known weaknesses. One
limitation in standard income analysis is that the family is treated as a black
box and that economic processes within the family are disregarded or treated
superficially. The study deliberately stays within the theoretical conventions
of standard income analysis, but moves beyond standard conventions in the
method of measurement by opening up the family black box. The
methodological questions are about variations in measurement within an
established theoretical framework.
The concept of well-being is derived from the consumption of goods.
Goods are bought in the market or produced in the family (or come through
transfers). In the translation of the goods avail- able to the family into
consumption available to family members, there is potentially both co-
operation and conflict. Through cooperation, value may be added; through
conflict, inequities may arise. The measurement of well-being should reflect
the result for individuals ‘after’ within-family production, co-operation, and
distribution.
The well-being of children depends on two processes, first how the
families in which children live fare in society, and then how children
specifically fare in their families. The study is organized accordingly; it
starts with the situation of families with children and then moves inside
families to the situation of children. Of course, there are children who do not
live in families and some of these may be among the most vulnerable of all
children, but the present study is limited to children who live in ‘normal’
family circumstances.
Well-being is estimated from the sum of income from sources outside of the
family (cash income) and the value added through within-family economies.
Assumptions
Data
The study draws on two types of data. First, the 1976 and 1986 Family
Expenditure Surveys, using data sets which have been harmonized for
comparability.5 Similarly harmonized later FES data were not available.
These data give socio-economic information for relatively large samples of
families, including cash income. The information on within-family
economies is based on time-use data from two surveys, the 1987 Social
Change and Economic Life Survey of six British towns (and their ‘travel-to-
work’ areas), and the BBC 1974–5 time-use survey.6 These surveys give
information on the use of time in paid work and household activities.
The income definition used is disposable family income (earnings from
work and assets, plus transfers from the state, less taxes and National
Insurance contributions). All monetary quantities are expressed in December
1986 £s.
Housework time is defined as time spent by adults on core housework
(e.g. cooking, cleaning), odd jobs (e.g. gardening, maintenance, DIY), and
shopping, but excludes child care.
The income surveys and time-use surveys are not undertaken with
overlapping samples. It has therefore not been possible to match data from
these two sources directly. Instead, the most likely pattern of time-use for
each family in the FES samples has been inputed on the basis of known
patterns of time-use for similar families in the time-use surveys. In this we
are obviously limited to working with those variables which happen to be
present and identically defined in both relevant data sets. The procedure has
been to estimate a predictive model of time-use based on the time-budget
data sets, separately for the 1970s’ and the 1980s’ observations, and to use
the resulting parameter estimates to time-use for the FES families. Appendix
B explains the imputation procedure in more detail.
Samples
Since our concern is with children and family economies, and in order to
keep the study within manageable proportions, the samples for analysis have
been limited to the ‘normal’ child/family range. Both young and old families
have been excluded, namely those in which the head of household is either
19 years or younger or 50 years or older. This greatly simplifies the analysis,
in particular the over-time comparisons in which the restriction to the
narrower age-band effectively removes the problem of differences in age
structure at the two points of observation. What is lost is information about
‘non-normal’ child families, but that is beside the main concern in this
particular study. Also excluded are households with more than two adults,
households headed by married couples in which one of the spouses is absent,
and in 1986 one household which is recorded with an exceptionally high
income. The latter exclusions are made on the basis of previous experience
with these income data sets which suggests that these smaller categories in
the sample may be exceptional or non-representative so that their inclusion
might distort the analyses.
This leaves families in the age range 20 to 49 years (age of head of
household), headed by a married couple or single person, with or without
children. ‘Marriage’ includes cohabitation, ‘children’ are under 16, or
dependent and under 19, the ‘head of household’ is the male parent/partner if
present, else the female.7
A remaining ‘exceptional’ category is that of single-father families. The
number of single fathers is in most cases too small for multivariate analysis
and some results for single-father families as a group are surprising in ways
that suggest that this may not be a very meaningful or homogeneous
category sociologically. Some results for single-father families are given in
the tables, but not much will be said about these results in the analyses.
Terminology
I use the term family rather than household since non-family households are
not included in the analysis (but will use household production and head of
household). By income is understood income from sources outside of the
family (cash income). Income enters the measure of well-being as
disposable income. The sum of disposable income and goods from
household production is referred to as consumption resources. The term
standard is used for equivalized values, so that income standard is
equivalized disposable income and consumption standard is the equivalized
value of consumption resources.
Robustness of results
FAMILIES
In 1986, mean gross income for all British households was £11,918 and
mean disposable income £9,708. Compared to 1976, gross income was up by
14.6 per cent and disposable income by 19.1 per cent (in real terms).
Households with children had a mean gross income in 1986 of £14,088 and a
mean disposable income of £11,545, up by 13.3 and 16.8 per cent
respectively.
Tables 3.1 and 3.2 show the distribution of gross income by source and
the level of disposable income relative to gross income for the sample of
families to which the present study is limited. (All tables are in Appendix
A.) While earnings are and remain by far the most important source of
family income, there are considerable differences in the composition of
income between family types and considerable changes over time.
Families with children, compared to families without children, generally
have a larger share of their income from state benefits and a correspondingly
lower share from earnings. The share of state benefits in total income
increases when the number of children is above two. Single-female-headed
families with children are strongly dependent on state benefits and other
transfers.
The main changes to observe from 1976 to 1986 are in the relative share
of earnings by heads of households (down) and in the relative share of state
benefits (up). The relative importance of state benefits specifically for
families with children increased quite dramatically, to about double. The
main non-change to observe is in the share of family income from wife’s
earnings in couple families, which is up only marginally.
In the trade-off with the state, as far as cash transactions are concerned,
families did very well during this period, receiving more in benefits and
paying less in taxes and contributions. Also, tax levels are down, bringing
disposable income up to a higher level relative to gross income.
Tables 3.3 and 3.4 show the distribution of time-use on average per day for
adults by the type of family and the number of children, in 1987 and 1974–5.
Couples for which time-use diaries have failed to be recorded for both have
been excluded. The observation covers 21 hours of the day, the period from
2 to 5 a.m. having been excluded for reasons of over-time comparability
since this period was left out in the 1974–5 survey. This adds up to 1,260
minutes (with a small margin of rounding-off error).
The pattern of time-use is roughly as follows, reading first the 1987
observations:
1. On average for all persons, irrespective of family type and children
(except for couples without children), men use more than twice as much
time as women in paid work, and women about twice as much as men in
housework and three times as much in child care.
2. Married women and men (all families considered) have about the same
patterns of time-use as all women and men.
3. Single men have a pattern of time-use which differs a good deal from
that of all men: less time in paid work and child care, and more leisure.
4. Single women have a pattern of time-use which is only moderately
different from that of all women: somewhat less in paid work and
housework, the same in child care, and some more leisure. Single women
with children use less time than other women in paid work. Single women
without children have a ‘male-type’ pattern of time-use.
5. In families with children, compared to those without children, women
spend less time in paid work and more time in housework and child care;
men spend about the same or marginally less time in paid work, significantly
more time in child care, and some more in housework.
6. For women, the pattern of time-use depends strongly on whether or not
they have children: with children significantly less time is in paid work and
more in housework and notably child care. For women who have children,
the number of children matters less for the pattern of time-use.
7. For men, the pattern of time-use depends less on children, except that
men with children spend more time on child care. Time in paid work
decreases marginally and in housework increases marginally with children
and the number of children.
All this is straightforward and intuitive, except possibly for the
observation that men’s time in paid work decreases with an increasing
number of children.
In the 1974–5 observation, the pattern of time-use has many similarities to
that summarized above but there are also some changes over the period.
First, the amount of time devoted to child care has increased. This is the case
for both women and men, and irrespective of family type and the number of
children. Women continue to put in much more time in child care than do
men, but the gap has narrowed, quite significantly so for married couples
(down from the range of 4 or 5 to 1, towards 3 to 1). Second, the distribution
of time in housework has changed. Women have moved to spending
marginally less time in housework and men considerably more. The addition
for men is greater than the reduction for women, so that in the aggregate
there is an increase in the time invested in housework. Women continue to
do more housework than men, but, again, the gap has narrowed. Third,
women, including married women specifically, have moved to more time in
paid work, although not much more, and probably not at all so for the
minority of women with three or more children. Men in all categories spend
less time in paid work. The increase for women is slightly less than the
decrease for men so that in the aggregate time in paid work has decreased
marginally. The observations on time in personal care and leisure may be
uncertain (because of the missing hours) but suggest for both women and
men less time in personal care, for both women and men without children
more time in leisure, for (married) women with children the same or
marginally less time in leisure, and for men with children the same or
marginally more time in leisure.
Overall, the two main features are a shift in time-use from paid work to
housework (in spite of the shift in the opposite direction for women) and a
considerable increase in the amount of time spent in child care, including per
child. The pattern of time-use has changed a great deal for both men and
women, and hardly less for men than for women.
In recent years, there have been very considerable changes in labour force
participation, and this is often considered to be among the most significant
recent social trends. While more women, notably married women, have
entered the labour force, the rates of economic activity of men have
decreased. The participation rates of married women have increased more
than the rates for men have decreased. For example, from 1973 to 1986,
participation rates increased for married women from 55 to 66 per cent and
declined for men from 94 to 89 per cent.8 The decline in male labour force
participation is reflected in the observations here in a decline in the
contributions of male earnings to family income and a decline in men’s time
in paid work. The stronger increase in the labour force participation of
married women is, however, not reflected in a similarly strong increase in
time in paid work or in the relative contribution of wives’ earnings to family
income. While aggregate participation rates have increased, aggregate time
in paid work has decreased. In married-couple families, the relative
contribution of male earnings to gross income declined from 76 to 68 per
cent while the contribution of female earnings increased only from 17 to 20
per cent. These seemingly contradictory results may be explained by
differences in wages for women and men, the (increasing) frequency of part-
time work for women, and other factors.
Consumption resources
Tables 3.5 and 3.6 show the sources of consumption resources by disposable
income and the value of women’s and men’s housework. (Remember that
housework does not include child care.)
1. On average in all families, two-thirds of consumption resources, as that
term is defined and measured here, is from income and one-third from
household production. In other words, adding household production, even
defined as conservatively as is done in the present exercise, increases the
value of available resources by upwards of 50 per cent.
2. Housework contributes more to total consumption resources in families
with children than in families without children. The added importance of
housework in families with children is the result almost totally of additional
housework by mothers.
3. In families with children, there is virtually no difference by the number
of children in the relative contributions of income and household production.
The presence of children matters strongly but the number of children hardly
at all.
4. In single-mother families, more than 40 per cent of total consumption is
from housework, which is a significantly higher proportion than the sum of
housework by mothers and fathers in married-couple families with children.
(But remember that single-mother families have the lowest absolute level of
income and consumption.)
In the period from 1976 to 1986, there was not all that much change in
these relationships. The relative contribution from cash income increased
slightly, as did the relative contribution from men’s housework, whereas the
relative contribution from women’s housework was slightly down. The
increasing importance of cash income might be unexpected since aggregate
time in paid work went slightly down, but is probably explained by higher
levels of transfers and lower taxes.
Effects of children
Consumption standard
So far, the analysis has been in terms of income and consumption per family.
We now turn to translating these values into estimates of standard after
adjusting for family size and composition. This is done by re-estimating
absolute values per family to equivalent values per family, thus producing
measures of standard which are comparable across families of different size
and composition.
Tables 3.9 and 3.10 show these estimated standards, relative to the mean
standard for all families. The main results are, reading first the 1986
observations:
1. Families with children have a lower than average standard, families
without children a higher than average standard, and these differences are
generally quite large.
2. Single-female-headed families have a lower standard than both
married-couple families and single-male-headed families. This is the case
among both families with and without children.
3. Among families with children, relative standards are in all cases lower
the higher the number of children.
4. When measured with the consumption standard, as compared to the
income standard, the differences between family types are in the same
direction, but of much less magnitude. Families with children are still below
average, but less so; families without children are still above average, but
less so; and single-femaleheaded families are still at the bottom of the
distribution, but at a higher relative position.
In 1976, these relationships were much the same: families with children
lower than families without children, single-female- headed families lowest,
and less difference in consumption standard than in income standard.
However, the observations also show some considerable changes over the
period:
1. The relative standard of families with children is lower in 1986 than it
was ten years earlier.
2. Among married couple families, the relative standard is about the same
in 1986 as it was in 1976 for those who are without children and with one or
two children, but lower for those with three or more children.
3. In single-female-headed families, the relative standard is much lower in
1986 than it was in 1976, both for those with and without children. These
changes over time are about the same in income standard and in
consumption standard.
Goods purchased in the market are, in a sense, the raw materials for the
family from which it generates consumption for family members. To these
raw materials, the family itself makes two additions. It produces additional
goods of its own, and family members co-operate in translating the goods
available to the family into consumption for family members.
The value of goods from outside of the family is set equal to family
disposable income. The value of household production is estimated from the
time invested in such production. The effects of co-operation is captured in
the equivalence scale, and with the absolute interpretation of equivalent
income, it is possible to estimate the value added through co-operation. Such
estimates are given in Table 3.11. The results are, reading first the 1986
estimates:
1. Disposable income per person is £3,515. Through the process of co-
operation, this is translated into an income standard per person of £4,744,
which represents a value added of 35 per cent. In other words, even if the
family produced no goods of its own, it would still, on average, add about a
third to the value of the consumption available to its members, relative to the
market value of available goods. This factor would obviously be higher in
larger families and lower in smaller families, and could be higher in
harmonious families and lower in families in conflict.11
2. Through household production, the family adds, as we have already
seen above, 56 per cent to the goods available to it from the market.
3. Taking into consideration the effects of both production and co-
operation the aggregate value added in the family is estimated to 113 per
cent on top of disposable income.12
These estimates of value added appear remarkably high. The family itself
produces more value than the value of the goods its income can purchase in
the market. Even so, the estimates are conservative. The equivalence scale is
modest and does not impute exceptional economies of scale compared to
what is conventional in the literature. Household production is valuated
conservatively by the housekeeper-wage method, and only housework
proper is included.13
The 1976 estimates are slightly different, giving higher estimates of value
added. This is a result of two factors, the increasing relative importance of
cash income (as we have seen above, in spite of reduced time in paid work)
and the transition towards smaller families. The latter has the effect of
weakening the economies of scale since available goods are shared between
fewer family members.
The change from 1976 to 1986 in income per person is akin to much-used
measures of economic growth, such as GDP per capita. This shows a rate of
‘growth’ of 31 per cent. That, however, is not a good measure of the trend in
consumption well-being since it does not take into account within-family
economies and changes in these. The results here suggest that per capita
income estimates of the standard of living strongly exaggerate the pace of
growth (in Britain during this period). The standard of living has increased,
but less so when measured after within-family economies than if measured
by per capita income.
1. Consumption resources have increased considerably less than has
income. This is a result of the declining relative importance of household
production whereby families add less goods through production of their own
relative to what they can buy in the market.
2. The increase in equivalent values is less than the increase in per capita
values. This is a result of the move towards smaller families.
The introduction of household production reduces the growth rate by 5
percentage points, and equivalizing reduces it by another 3 percentage
points. All in all, this brings the estimated increase in standard of living
down from 31 per cent when measured in per capita disposable income to 23
per cent when measured in the value of consumption available to persons
after within-family economies. This difference could be taken as a measure
of the ‘cost’ to society of the changes that have taken place over this period
in family arrangements and the economies of the family. For example, the
‘cost’ of the reduction in average family size alone is in the magnitude of 10
per cent of economic growth as conventionally measured.
There is no mystery or statistical magic in any of these results. The costs
in question are very real costs. If relatively less work is done in the family,
relatively more goods need to be purchased in the market, for example more
processed or ready-made food or more paid help. If average family size goes
down, the population will need more houses, more cars, more washing-
machines, stoves, TV sets, and so on, and will hence have less consumption
per person from given aggregate resources.
These changes in family arrangements may to a greater or lesser degree be
the result of voluntary choices, in which case the estimated ‘costs’ could be
regarded more or less as another way of taking out increasing prosperity. We
may be buying more goods in the market, and doing less housework,
because we can afford to. Families may be becoming smaller because young
persons can afford to establish families of their own earlier and elderly
persons are able to manage longer on their own. No doubt, other factors are
also in play, some of which are clearly involuntary (ageing of the
population) and some of which may be more or less involuntary (divorce,
single motherhood, the family effects of increasing female labour-force
participation).
PERSONS
Within-family inequity
Table 3.12 gives estimates of relative income and consumption standards for
adults and children, under the assumption of equity within families. This
works so that in any given family, adults and children are recorded as having
the same standard; each person is allocated the equivalent income and
equivalent value of consumption resources, as estimated from family
income/consumption resources.
In spite of the assumed equity within families, it turns out that the
standard of living is different for children and adults as groups. These
differences do not result from any discrimination against children in their
families, but from the position of families with children in the overall
distribution of income. Results:
1. Children have a lower standard of living than do adults. In 1986, the
average income standard of children was 75 per cent of the average for
adults.
2. Children also have a lower standard of living than do parents, but this
difference is much less than between children and all adults. For example,
children in married-couple families have an income standard which is 94 per
cent of that of married parents. The difference is, however, persistent. The
average for all children is lower than the average for all parents, that of
children in married-couple families lower than that of married parents, and
that of children in single-mother families lower than that of single mothers.
3. The difference between children and adults is less in consumption
standard than in income standard. The difference between children and
parents is about the same in consumption standard as in income standard.
4. From 1976 to 1986, the standard of living (income standard as well as
consumption standard) improved less for children than for adults and slightly
less for children than for parents. Both children and adults had a slightly
better development in income standard than in consumption standard.
5. Among children, the relative standard is slightly higher than average for
children in married-couple families and considerably lower for children in
single-mother families. Children in marriedcouple families had a slightly
better development from 1976 to 1986 than the average for all children;
children in single-mother families lagged further behind compared to other
children.
6. Along the family life-cycle, the relative standard of children improves
with the age of their parents.
7. Among adults, the standard is much lower for those who have children
in the family than for those who are without children (but less so in
consumption standard than income standard), and those with children had a
slightly slower increase in standard of living than those without children.
Conclusions
Three stories emerge from this study. There is a story about the family, a
story about children, and a story about measurement.
The family
Production within the family adds very considerable resources to the goods
families are able to buy in the market with available cash income. Using
conservative definitions and methods of valuation, household production is
found to add as much as just over a half to disposable income for all families
and about two-thirds for families with children.
Production of goods is only the first job the family does in respect to the
consumption available to family members. It also translates its aggregate
bundle of goods (be they bought in the market or produced in the family)
into consumption for individual family members. In this process the family
does a second job through co-operation. This (normally) enables family
members to each have a higher value of consumption than they could each
have had on their own from the same aggregate bundle of goods. The degree
of co-operation is without doubt very different from family to family, but a
family would not be a family if there was no co- operation (although in
extreme cases of destructive family conflict there may be ‘negative co-
operation’). With reasonable assumptions it can be suggested that the
average sum value of consump- tion available to family members ‘after’ co-
operation is between 30 and 40 per cent higher than the per capita market
value of the consumption resources available to the family. The value added
through co-operation depends strongly on the number of family members
and will be higher in families with children than in families without children.
These estimates should not be taken as precise statistics, but rather as
indications of the order of magnitude. It is clear enough, however, that in
terms of consumption well-being, cash income and the goods cash income
can buy in the market are only the raw materials that go into final
consumption. The average family adds as much or more value on its own as
it can bring in from the outside. This is not to say, obviously, that the family
could make do without the goods and services it buys—it would have
nothing to process further through its own activity—but it firmly establishes
the family as a substantial institution of production.
Families are sensitive in their economic arrangements to the presence of
children. Mothers are much less likely than other (married) women to have
paid work, or, if they have paid work, are likely to work less. Both mothers
and fathers devote more time than other adults to housework and (obviously)
child care, in particular mothers. Because of these influences, families with
children have less cash income than do families without children, other
things being equal. The ‘opportunity cost’ of children depends very much on
the age and number of children, and on the ‘age’ of the family in the family
life-cycle. As a rough approximation for young families with young
children, the opportunity cost may be in the range of 30 per cent of expected
income, expected meaning other things being equal but no children. (This is
‘opportunity cost’ only and does not include outlays to cover the cost of
child-rearing.) Hence, although consumption value is added through family
economies, families with children on average start from a lower income
base.
Families do not take the loss of income sitting down. They adjust (i.e. in
the main mothers adjust) by increasing the production of goods within the
family. The loss of income is to some degree compensated by additional
household production. This reduces the opportunity cost but far from
eliminates it. For young families with young children the value of
consumption resources (from the mar- ket and household production) is in
the order of 20 per cent less than the expected value without children.
These effects of children on the resources available to their families carry
through to differences in the standard of living of families with and without
children. The standard of living in families with children, whether measured
by income or total consumption resources, is considerably lower than the
average standard of all families, and is lower the higher the number of
children.
In addition to the level of consumption resources, families with children
differ from other families in the composition of resources by source. Not
only are families with children relatively more dependent on household
production, they are also in terms of cash income relatively more dependent
on state benefits. This is as one would expect since some state benefits are
aimed directly to compensate for the cost of child-rearing. However, state
benefits are nowhere near enabling families with children to maintain a
standard which is on a par with the average standard in the population.
Over the period covered in this study—1976 to 1986—there were very
considerable changes in these aspects of family economies. In time-use,
there was a movement ‘into’ the family, with more time invested in
housework and child care and less in paid work. In terms of cash income, in
families with children, the relative contribution of earnings went down while
the relative contribution of state benefits increased quite strongly. The value
of household work relative to cash income decreased marginally and the
value added through co-operation was reduced substantially. Although
economies of scale in the family are assumed to have remained unchanged,
their impact was reduced because average family size was reduced. This
movement towards smaller families brought the value added down by 2 to 3
percentage points. In order of magnitude, this represents about 10 per cent of
economic growth as measured by the conventional per capita income
method.
All this adds up to a decline in the relative standard of living of families
with children. In 1986, these families were further behind the average
standard of living in the population than had been the case ten years earlier.
The brunt of this relative decline was carried by single-mother families,
while married-couple families roughly maintained their relative position.
Do these analyses suggest that the family has become more important or
less important as a basis for economic well-being? Only in the sense that
families have become smaller is there a sign of ‘decline’ in the importance
of the family; in all other respects the family seems to have become more
rather than less important. The direction of shift in economic activity is from
market work to household work. The direction of shift in income
dependency is away from market earnings. The direction of shift in the
opportunity cost of children is towards increasing opportunity costs in terms
of cash income, while the opportunity cost in terms of consumption
resources has remained stable. This is interpreted as families having become
increasingly ‘vulnerable’ to children in terms of income but that this
increasing vulnerability is neutralized by compensating adjustments in
within-family activity.
Children
Measurement
APPENDIX A: TABLES
TABLE 3.1 Income by source, 1986
TABLE 3.2 Income by source, 1976
TABLE 3.12 Income standard and consumption standard, children and adults
TABLE 3.13 Inequality under alternative assumptions about within-family
distributions
0.7–0.5 scale gives the 35% estimate. A flatter scale would result in a lower
estimate of value added and a steeper scale in a higher estimate. For
example, a scale of 1.0–0.8–0.6 would give the value added at 24%, and a
scale of 1.0–0.6–0.4 at 48%. This is a bit different from the similiar results
reported in Ch. 2 because the analysis here is on a more restricted sample.
12 With the alternative scales given in note 11, this estimate would shift to
95% and 135%, respectively.
13 However, the value added is estimated relative to disposable income,
which excludes non-cash resources from outside the family, e.g. ‘free’
services from government or ‘fringe benefits’ from employers. This may
have the effect of exaggerating the relative importance of within-family
economies.
14 Cf. e.g. Young 1952.
15 Brannen and Wilson 1987; McRae 1987; Pahl 1989; Vogler 1989.
16 Sen 1984; Harriss 1990.
17 Glendinning and Millar 1987.
18 Piachaud 1982; Dwyer and Bruce 1988; Lazear and Michael 1988;
Haddad and Kanbur 1989; Jenkins 1994.
19 For example, Davies and Joshi 1994.
20 e.g. Dilnot et al 1984.
21 Harriss 1990.
22 Brannen and Wilson 1987; Thomas 1991; Hoddinot and Haddad 1995.
23 Easton 1986.
24 These tables are not reproduced.
25 This result is consistent throughout the study: there is less difference
1. Sex
2. Age group
3. Age of youngest child
4. Number of children
5. Employment status (i.e. full-time, part-time, not in the labour market)
6. Paid work time
Definitions must match for all the variables listed above. The first four are
straightforward. Employment status reflects full-time/part-time conditions of
employment, and is not directly calculated from actual or usual hours
(though it will correspond very closely with the latter). This can be matched
in the FES. Paid work time is more problematic. The diary work time is
accurate work time for the period of observation and includes travel-to-work
time. The best we can get in the FES is reported usual hours. The
relationship between these is not extremely close, as hours reported (rather
than recorded) may be notional or contractual hours (see Robinson and
Gershuny 1994). However, the mean levels for FES reported hours and time-
budget diary hours are very close, once respondents with zero scores are
removed.
A number of other available variables have been tested and excluded as
they were not significant. Conversely, some significant variables have been
excluded because they are not available over all files—for instance, income
source (but, importantly, not income level, which has very little effect on
time-use).
The simple model performs relatively well, compared with fuller models
which were estimated on the 1987 SCEL data, and which included detailed
information about the effects of income amount and type, and wage level, on
time-use.
1 − 10% if i is a child
τ = {
1 + 10% if i is an adult.
Σi πiis equal to Y, the family income, since Σiρi, equals one. If we express
the actual share achieved as a proportion of a perfectly equal distribution, we
find this varies with the family composition. The actual share can be defined
as:
βi = ρi × N ,
where N is the number of persons in the family. For a ‘bias’ of 10 per cent,
in a one-child/one-adult family, this will be:
1+10% 1.1
β1 = 2 × = 2 × = 1.1
(1+10%)+(1−10%) 2
1−10% 0.9
β2 = 2 × = 2 × = 0.9
(1+10%)+(1−10%) 2
for the adult and the child respectively. For other family compositions the
figures move away from 10 per cent but not extremely so. Table 3.17 gives
the proportional shares in families with one or two adults and one to five
children. Children’s shares vary from a low of 87 per cent (single child with
two adults) to a high of 96 per cent (five children with a single adult). It may
seem counterintuitive that a group of many children will do better than a
single child, but bear in mind that (i) in a large family each child will be
competing against other children as well as against adults, and (ii) large
families will tend to be worse off at the level of equivalent income.
This strategy improves over the technique of simply reducing children’s
income by a fixed proportion in that it takes account of family composition,
and in having an interpretation in terms of relative ability to demand shares
of consumption.
Table 3.17 Shares of the cake where children suffer a 10% ‘bias.’
PART II
Reform
INTRODUCTION: SOFT MEANS FOR
HARD GOALS
When the necessaries of life are once provided, everyone should seek to
increase the beauty of things in his
possession rather than their number
or their magnificence.
Alfred Marshall, Principles of Economics
range of
purposes—and are used elsewhere in this book, although mainly
for convenience as a way of summarizing
distributional information—but
are, paradoxically, not very useful as a basis for comparative conclusions on
‘true’ extents or trends in poverty.
7 Atkinson and Michlewright 1992.
8 Erikson and Goldthorpe 1992.
9 Goldthorpe 1995: 10.
4
CLASS
INEQUALITY
ANALYSIS
ANALYTIC RESULTS
Conclusion
There are two results, one of stability and one of change. Both the analysis
of actual data and the
simulations show measures of asso- ciation and
measures of distribution to behave differently to the same
changes in the
social competition.
Both analytic results are correct; there is at one and the same time stable
association and changing
distribution, stable difference of treatment, and
changing inequality of distributional outcome. The question
is not which
finding is the correct one, but what this combination of stability and change
means for
the question of class inequality.
In the face of this evidence, the proposition of stable class in- equality
falls. There is no ambiguity about
this conclusion accord- ing to the primary
position on the concept of inequality. If there is
ambiguity in the concept of
inequality, the proposition nevertheless falls
according to the minimal
condition.
Interpretation
REFORM
which is
a log transformation of the OR. Its numerical value is simply more
attractive than the unwieldy raw ORs. In all
references to association, we
compare the two extreme classes, higher and lower. ? is one-fourth of the
natural logarithm of the OR, when four frequencies are used in the
calculation.
13 The analysis of inequality in Table 4.1 takes the class of the individual
INCOME
MARGINAL REDISTRIBUTION
One of the difficulties which arises from the static quality of the standard
method is that redistribution is
estimated from the profile of the tax/transfer
system as a whole. Of course, we do not keep constantly starting
afresh
with a virgin distribution of income upon which a system of transfers and
taxes is imposed. At any point
in time, we start from prevailing
distributions and policies and then introduce, if we so choose, some policy
change. In a dynamic perspective, redistribution is captured more
realistically in marginal effects on
inequality of marginal policy changes
than by estimating, more theoretically, global effects of entire policy
systems.
In Britain, the distribution of income moved in the direction of equality
from around 1950 to the mid-1970s;
that trend was then broken and the
distribution started to move back towards inequality. By the mid-1980s, the
1950 distribution had been restored and towards 1990 it had become yet
more unequal. This shift in the
distribution of income offers a sharp test on
the politics of inequality: if there
is a link between policies and distribution,
we should expect any shift in the distribution, including towards
inequality,
to be associated with changes in policies.
Atkinson, in addition to documenting the trends, has analysed the
movement towards inequality.5 He has pointed to three factors. First, a shift
from
work in the sense that fewer households are headed by persons in
work (down from about 70 per cent in 1970 to
less than 60 per cent in
1985) and a smaller proportion of household income is from work (down
from more than
80 per cent in 1975 to just over 70 per cent in the mid-
1980s). This factor, he finds, accounts for about half
of the increase in
inequality from 1975 to 1985. In addition, the income differential between
the in-work and
the not-in-work groups increased, notably after 1985.
Second, a rise in earnings inequality among those in
work. Around 1975,
among full-time workers, the earnings of those in the top decile were about
three times the
earnings of those in the bottom decile, whereafter the ratio
increased steadily to about 3.4:1 in 1990. Third,
the distribution of income
among those not in work. During the 1980s, while the value of non-state
pensions
continued to follow earnings trends, the relative value of state
pensions and benefits declined when the basis
for upgrading changed from
earnings to prices, the result being a relative deterioration in the income
position
of those depending most heavily on state pensions and benefits.
The story is complicated, with a range of economic, demographic, and
political factors contributing to the end
result. Policies clearly play a role.
Among the relevant policy factors is that of changing rules for the
upgrading of social security benefits. This accounts for a good deal of the
increasing inequality among those
not it work, and for some of the
increasing difference between those in work and those not in work. How
much
exactly this factor contributes to the total explanation is not clear, but
for the present purpose it is enough
to say that it is significant. The new
British trend towards inequality cannot be put down to ‘automatic’
forces of
long-term stability in the distribution of income; it is to some considerable
degree induced by
changes in policies. The change from earnings to prices
as the basis for upgrading state benefits was seen,
when it was made, as a
marginal adjustment. Over time, the consequences have
accumulated to a
very significant effect on inequality in Britain. This was possibly not
anticipated, and this
experience would then seem to demonstrate in a rather
illuminating way that the distribution of income is
indeed sensitive to even
marginal adjustments in policy.
In a comparison of income inequality trends in five countries during the
1980s, Fritzell found stability in
Germany and Canada and a trend towards
increasing inequality in Sweden, Britain, and the USA.6 Redistributive
policies are found to contribute to
explaining both stability and change. In
all five countries, an increase in factor income inequality was
observed; in
Germany and Canada this was counterbalanced by changes in policies
which pulled in the opposite
direction, hence final stability; in Sweden,
Britain, and the USA, changes in the distribution of factor income
carried
through to the distribution of disposable income, and redistributive policies
were reshaped so as to
add to this effect, rather than, as in the two other
countries, to neutralize it.
In an earlier study, I compared income redistribution in Norway and
Sweden, using carefully co-ordinated survey
data for 1982.7 That
comparison turned out to
have a quasi-experimental flavour: while factor
income was identically distributed in both countries,
disposable income was
more egalitarian in Sweden than in Norway. This suggests that the
explanation of the
difference in the end distribution lies ‘between’ factor
and disposable income, i.e. in political forces which
redistribute factor
income ‘afterwards’, rather than in economic forces which determine the
original
distribution of factor income (although some uncertainty about
second-order effects on factor income remains).
There also turned out to be a simple difference in the structure of
policies. In Sweden, transfers as well as
taxes were on a higher relative
level; in Norway, both had a more progressive distribution across income
groups. Through cross-simulation, the Swedish levels of transfers and taxes
were imputed into the Norwegian
system while the Norwegian
distributional profiles were maintained, and the Norwegian profiles were
imputed
into the Swedish system while the Swedish levels were maintained.
Both simulations, i.e. including the one in
which only the levels of transfers
and taxes were manipulated, had the effect of changing the distribution of
disposable income, compared to the pre-simulation distributions. As in the
actual experience in Britain, this theoretical simulation again shows that the
distribution is indeed sensitive
to marginal changes in policies, in this case
also to marginal changes in the levels of transfers and taxes
within constant
profiles.
STRUCTURAL FACTORS
LIFETIME REDISTRIBUTION
CONCLUSION
were
successful in redistributing income, there could be costs in the form of
inefficiencies which could make the
policies prohibitive. Such costs could
arise through behavioural responses (e.g. work inefficiency) or through
attitudinal responses (e.g. cultures of dependency). I have considered such
‘moral hazards’ elsewhere (Ringen
1987, 1995). Although the matter is
complex, the following seems to be a good rule of thumb. By ‘cost’, I
understand ‘welfare loss’. (A budget expenditure is not in itself a cost to
society.) There are costs; cost-free
redistribution would be too good to be
true. Therefore, the cost side would have to be considered in the
formulation of policies. However, aggregate social costs are often low, for
example because income and
substitution effects balance each other out. (I
would not accept a definition of cost which includes only
substitution
effects.) Likewise, while some policies may encourage dependency, others
may stimulate
self-reliance. If costs are unavoidable, policies can be
designed so as to reduce potential costs. Nevertheless,
there may be trade-
offs between redistribution and social efficiency. Although cost-free
redistribution is
unlikely, effective redistributive reform (which is about
modifying inequality and not about creating equality)
is possible without
prohibitive costs.
5 Atkinson 1993.
6 Fritzell 1993.
7 Ringen 1986.
8 Uusitalo 1989, and personal communication on trends from 1985 to
1993. In the
most recent observation, from 1992 to 1993, there has been a
modest movement in the direction of inequality,
which may or may not be
the beginning of a new trend.
9 Hills et al. 1993.
10 Harding 1992.
11 Falkingham et al 1993, using a model population rather than a cohort.
12 There are four situations in which policy intervention for the
equalization of
income would unquestionably be correct: (i) the existence of
poverty which could be reduced or eliminated through
redistribution; (ii)
the existence of indefensible pockets of relative deprivation which could be
reduced or
eliminated through redistribution—relative deprivation among
children would be a case in point; (iii) the
existence of degrees of inequality
which are objectionable to community or threatening to cohesion or
efficiency;
and (iv) strong and rapid shifts towards inequality, or threats
thereof.
6
At the end of 1989 and the beginning of 1990 I found myself in Chile,
immediately after the successful
congressional elections and during the
days when the Chileans were finally able to feel confident that
democracy
would be restored. We who have grown up in Western Europe after the
Second World War have never
experienced the loss of freedom. That is our
good fortune, of course, but we should also be aware that we do
not know
the physical reality of oppression. This was brought home to me during a
long evening in the
company of a dignified Chilean, an elderly lady whose
husband had disappeared during the dictatorship, whose
son had been killed
by having his throat cut in broad daylight in central Santiago, and whose
daughter had been
in exile until the year before, and who was now herself
elected Diputado for the Partido por la
Democracia, one of the parties in
the Concertacion which beat the Pinochet crowd in the
elections. As she
told her story, in a soft-spoken way which was unreal to me, this visiting
North European
learned something about democracy, and in particular the
danger of its absence.
In The Possibility of Politics, I tried to analyse the welfare state in such a
way as to comment on
the feasibility of piecemeal, step-by-step, reformist
policies. In intellectual circles, reform is suspect as
seen from both left and
right. I was therefore delighted, being a middle-of-the-road man myself, to
be able to
pass a favourable verdict on the power of reform.
I have always associated political reform with political democracy. I
believe that democracy generates reform
because it will not be accepted to
let problems lie where they fall. I also believe that democracy limits
politics
to reform since majorities will accept revolutions to democracy but not
revolutionary shifts within
democracy. In a sense, therefore, the analysis I
was conducting was about reason in the muddled arrangements of
the
mixed-economy democracy. I did not feel, however, that I could extend the
analysis at that time to the issue of democracy. ‘If it were not’, I wrote in an
introductory section,
‘that I already have more than a reasonable share of
broad questions on my plate I might have been tempted to
add, Is
democracy possible?’
I resisted the temptation then, wisely I’m sure. Today, however, I intend
to give in and to take up the
question I then felt I had to set aside. After all,
this is an occasion on which it is accepted—perhaps even
encouraged—to
address issues which would in other circumstances be frivolous. So, then,
does democracy work?
I have a professional interest in the application of social science to issues
of concern in society. Applied
research is intended to contribute to
something practical. What social research contributes to is perhaps not
always clear. I want to argue that we have a role to play in the workings of
democracy and that a programme for
applied social science can be
formulated in terms of fulfilling this role.
The last few years have seen revolutions of democracy throughout the
world. I have in mind, obviously, Eastern
Europe, but also the Phillipines,
Argentina, Zambia, South Africa (hopefully), Nicaragua, and certainly
Chile—once the democratic pride of South America, then the world’s most
despised dictatorship, and now again a
working democracy. I think of the
fall from grace of the outlandish idea in Africa that one-party rule is
necessary because of African tradition. I have in mind the eradication
(almost) of military rule in Latin
America and the many attempts, although
some unsuccessful so far, in the direction of democracy in, for
example,
China, Nigeria, Burma, Bangladesh, Kenya, Côte d’Ivoire, Algeria, and not
least the tragedy of Haiti,
where the first free elections after almost 200
years of independence were soon enough set aside. Having
visited that
country several times and been captured by the cultural richness of a proud
people suffering no
end of political corruption and material deprivation—
this is a land where the very physical soil has been
plundered—I feel that I
have experienced the degradation and destructiveness of bad government in
its extreme
form. Those who need to see for themselves the price of non-
democracy in the elementary life chances of
ordinary people, let them go to
Haiti.
The logic of applied research is that the researcher puts to use his or her
tool-box for resolving some problem
which arises from the conduct of
practical affairs. There are odd problems which serious scholars would not
be bothered with and there are wicked problems which scholars of integrity
would not touch, but generally, in applied research, we take on problems
such as they are presented to us.
In the social sciences, problems are often presented to us by ‘decision-
makers’. That is as it should be, but
as a programme it is too narrow. The
act of decision-making, certainly in a democracy, is only the final step
in a
long process which leads up to the moment of truth when it is decided to do
A and not to do B. A proper
ambition for the social sciences is to respond to
the whole range of problems which arise in this entire
process and not only
to those of the final stage. The clients of applied social research are not only
decision-makers, but rather, and primarily, the public—the women and men
who participate in politics as
citizens. We should take up not only those
problems which are articulated by decision-makers but any problems
which
present themselves from the process itself, however they may or may not be
articulated. We should work
for the political process rather than for the
politicians.
The democratic revolution is enormously positive, but also chilling. The
most troubled peoples of the world are
turning to democracy for solutions
to their problems. I am quite convinced that they are right in so doing, but
the burden the democratic system of government is now asked to carry is
staggering. More is expected than can
be delivered. The coming of
democracy to the former Soviet Union is not exactly celebrated as a
triumph. In
Czechoslovakia, the Velvet Revolution tore the Federation
apart. In the first free elections in Poland in 1991,
and likewise in Zambia,
only about 40 per cent of the electorate chose to use the right to vote which
they had
just won.
In the established democracies we may feel safe, but I suggest we should
not be complacent. It has happened
before that democracy is lost. We
cannot take it for granted that our democracies are strong enough to stand if
the democratic experiments around us fail. On the European scene, some
politicians seem to believe in
supernational power without democratic
institutions. ‘[D]emocracies have always been, and still are, failure
prone.
They were short-lived, and by all accounts ill suited for survival.’ Thus we
are reminded by Giovanni
Sartori of the historical experience. Horrors we
thought were confined to the scrap-heap of history in our part
of the world,
have proved themselves to be alive and well. We need at this time to
be
deadly serious about the importance and difficulty of democracy and to give
careful thought to its meaning
and practical mechanics.
DEMOCRACY
SCIENCE
In 1936, Ragnar Frisch, the pioneer in the theory of planning, who was,
with Jan Tinbergen, to be the first
Nobel Prize winner in economics, wrote a
small memorandum which he entitled Den prinsipielle sondring mellom sak
og vurdering. There is a richness of nuance in this
Norwegian title which it
is difficult to capture in English, so when I offer The theoretical distinction
between fact and value as an English translation I am not doing full justice
to the original.
Professor Frisch was concerned with the relationship between science
and politics and with what social science
could do for politics. He
proceeded to define this relationship around the distinction between fact
and value.
Values enter science, he found, through the assumptions upon
which analysis rests. Value problems can and
should be resolved in
assumptions. The remaining problems are then factual, and scientific
analysis proper can
start. Once assumptions are given, it is possible,
although difficult, to achieve objectivity of analysis in
the sense that results
properly derived must be accepted by all who are competent to understand
the research.
The politician also, still following Frisch, confronts both types of
problems. The choice of goals is a value
problem, as is the decision about
which means are legitimate. Once this is established, however, the choice of
means among those which are allowed is a problem of optimality which can
be resolved through analysis.
From this Frisch went on to suggest what amounted to a division of
labour between politics and science.
Politicians should take care of the
value element, they should set goals and establish the boundaries for what
can and what cannot be done. The experts should then step in, take the
value premisses handed to them by the
politicians as their assumptions, and
find the most effective means.
Frisch stuck to this model throughout his life and never lost his
confidence in the ability of science to find
the right means once goals were
given. But he became increasingly impatient with the politicians who never
seemed able to do their part of the job. In his despair he constructed an
elaborate questionnaire which he
intended to administer to the members of
Stortinget, the Norwegian parliament, and through which he
would force
them to choose a consistent set of goals so that he and his colleagues could
get to work on their
computers and find the best means. This project never
came to fruition, and no doubt that was just as well.
Any social science student of my generation who suffered years of agony
at university under the wretched
‘positivism debate’ will shy away from
Frisch’s distinction between fact and value, but that is a trivial
matter which
there is no need to brood over here. Frisch wrote his memorandum when
he
was asked by a group of leading men in the business community to direct
an ambitious study of ‘economic
structures’. There had been an element of
expert rule thinking in the deliberations in this group and perhaps a
suggestion that the economic conditions of the nation might have been
better had expertise been trusted more
and politics less. Such ideas were
hardly surprising given the impotence of political authorities at the time
to
face up to the overwhelming problems of stagnation and unemployment,
but any hint of superpolitical
authority was, of course, an extremely
sensitive matter in Europe in the mid-1930s. Frisch was a radical in
politics
and he felt a need to formulate a position in which he disassociated himself
from any suggestion of
expert rule. With our eyes it might seem that he
handed over half of the political process to the experts, but
in the context of
the time, what he did was to defend the primacy of politics over expertise.
As all first-year students know, the strict distinction between fact and
value is not philosophically tenable,
and the idea that the value problem in
the social sciences can be cleared away by being confined to assumptions
had been rejected at the time Frisch wrote his memorandum, notably by his
colleague in Sweden, Gunnar Myrdal,
in a dissertation published in 1929
(and republished in revised form in English in 1953 under the title The
Political Element in the Development of Economic Theory). Those who
want to criticize my hero on this
point will have a field-day, but they will, I
think, miss the essential. By suggesting a division of labour
between
politics and science, Frisch claimed not only that there are certain problems
which only science can
solve and for which scientists should take
responsibility, but also that there are problems which science
cannot solve
and which scientists should therefore not take on. Exactly where to draw the
line of demarcation
may be in dispute, but that there is a line was right then
and remains right today.
All forms of political decisions are political, be it the choice of goals or
the use of means. No line of
demarcation there. There is, nevertheless, a
factual element in politics which is, in principle, non-political
and upon
which rational political action depends. This element is information. For
citizens to
participate in politics, they must be informed. Before political
decisions can be made, there must be
information on which to act. Having
abandoned the line of demarcation between fact
and value, I suggest an
alternative line drawn between information and decision; between, on the
one hand, the
questions of how things are and what can be done, and on the
other hand, the question of what
should be done. Let me, further, push to
the background the politician and pull to the foreground the
political
process. Let me suggest information as the common ground on which
science and politics can meet
without scientists playing politicians or
politicians playing scientists; information is what politics needs
and science
can provide. And let me, on this basis, and otherwise in Frisch’s spirit of a
division of labour,
try to redefine the job society should expect the social
sciences to do.
First, as social scientists we should accept responsibility for informing
the democratic process. We should
accept this responsibility (although
shared with, for example, journalists and educators) because we are the
custodians of the methodology which is best suited for the job of
establishing how things are and what can be
done. As has been pointed out
by Sten Johansson, the political methods of discourse, struggle, conflict,
and
compromise are useless for resolving questions such as: Is public health
improving or deteriorating? Is crime,
violence, and child-abuse on the
increase? What are the effects of changes in social security legislation on
poverty? How widespread is the problem of stress in the workplace? Do tax
cuts stimulate effort and initiative?
No debate, however sophisticated, can
produce reliable answers to questions such as these. These are empirical
problems and empirical problems should be resolved by analysis. To great
benefit for enlightened politics, we
do have generally accepted ways of
answering some persistent questions of this kind, for example about
unemployment and inflation, but we need authoritative information on a
much wider range of social and economic
trends.
I want it to be added that there should be no delusion about the limits of
analysis for producing irrefutable
answers to empirical questions. The
answers we produce will always be marred by uncertainty and there will
always be more questions asked than answers given. What I believe we can
and should say, however, is that there
is no alternative method to that of
careful analysis which can do the job better. Had we wanted, in 1957, when
Harold Macmillan told the British people, ‘You’ve never had it so good’, to
ascertain whether he was right, we
should not have organized a round table,
we should have consulted the statistics or
started a research project.
Second, we should not only accept responsibility for informing the
political process, we should claim this as
our authority. The political
struggle does not stop at the domain of information. The concern of the
politician
is, rightly, to be elected or re-elected, to win power or to stay in
power. Politicians who can doctor the data
to their advantage will do so, and
it is in the nature of competitive politics to spread disinformation. Since
we
cannot outlaw such behaviour without outlawing democracy—indeed, since
strong democracy depends on
ruthlessly aggressive politics—there is a need
to assure that the political process is independently informed.
Decisions
taken from distortions will be the wrong decisions, but much more
importantly, citizens’ control over
politics is not possible unless citizens are
informed. Hence, I do not agree with the great Henrik Ibsen who,
when
challenged, in a poem in 1875, wrote, ‘I prefer to ask, it is not my calling to
answer.’ Our calling is,
precisely, to answer.
In Britain today, government agencies retain ultimate control over the
publication of research results from
projects they have contracted. In the
social sciences we are dependent on government contracts and we therefore
do not have freedom to publish our results. This is a miserable state of
affairs which should be brought to an
end. It should be for the scientific
community to decide, through its system of peer review, what and how to
publish, and the power of the purse should make no difference to this.
Contract research for government
agencies is on behalf of the public and
the public should know what the research finds. It does not matter that
difficulties over publication usually do not arise, because it matters very
much in the few cases when
difficulties do arise—and there is always the
possibility of self-censorship.
Third, although we should claim the domain of information as ours, we
should, as social scientists, not
transgress into the domain of decisions. As
citizens we have the same rights as other citizens, but as
scientists we have
no special rights. We have some expertise with regard to the questions of
how things are and
what can be done, and therefore responsibility to serve,
but we have no competence on questions of what should
be done, and
therefore no special right to influence. If we use our expertise to inform the
political process,
we strengthen democracy on one of its central
assumptions, that of an informed
citizenry. If we claim authority which we
do not have in order to gain influence, we weaken democracy on one of
its
other assumptions, that of a decision-making process without distortions. It
is easy to be critical of
others who use unfair means to win influence, for
example money. We need to be equally critical of our own
ways. The
academic community is not a nobility with special privileges. We should be
activists, certainly, and
when necessary prepared to confront politicians in
the most forceful manner, but activists on the basis of
legitimate authority.
As social scientists we have no other expertise than information to offer; we
cannot do
surgery, we cannot build bridges, we cannot explode bombs.
Personally, I feel that the job of information is
important and sufficient, but
even if I did not, I should be made to understand that I have no business
trying
to convince the world that I am an expert in the rights and wrongs of
political decisions. Hence, I do not
agree with the great Karl Marx who, in
his Theses on Feuerbach in 1845 said, The philosophers have
only
interpreted the world; the task is now to change it.’ Our job is, precisely, to
interpret the world, and, precisely not, to change it.
A good society is a society in which citizens can live good lives. The good
life is subjective to the
individual, but rests on objective conditions for its
pursuit. These conditions are the domain of politics. The
subjectivity of the
good life is the responsibility of free citizens and is outside the business of
the state.
The conditions which make the good life possible are liberty, law, and
community. Liberty is the availability
of free choice, including reasonable
resources and options for choice in lifestyle. Richness is trivial.
Community
is social guidance for wisdom in choice.
Good society depends crucially (but not exclusively) on good
government. Good government is necessarily
democratic. Democracy is
superior to autocracy in respect both to justice and to efficiency.
In democracy, the state has duties towards citizens. These include the
safeguarding of liberty (from overall
security to minority rights); the ability
to enact and execute adequate law; and the distribution of wealth so
as to
make choice a universal reality.
Equally, citizens have duties to each other. These include to make use of
liberty, and when participating to do
so with integrity; to co-operate in
community; and to comply with decisions of the state properly made.
Contradictions between liberty and duty, liberty and law, or liberty and
community are possible, but not
necessary and intolerable as long as
democracy is sufficiently vibrant to temper absolutism in the
interpretation
and practice of duty, law, and redistribution.
REFERENCES