others and the whole system. An aggregate index of well-being is hence worthless, for it would
hide the information given by fluctuations of the system13.
Anyway, in CFM we face the aggregative problem at a lower level, since we collapse the
indicators in a more general dimension of well-being, i.e. the achieved functioning. We in fact
move «from the space of elementary indicators to the overall evaluation of a given functioning
for each unit of analysis» (Chiappero Martinetti, 2000:7). According to Sen the capability
framework allows great freedom in choosing the suitable aggregative strategy: «Quite different
specific theories of value may be consistent with the capability approach, and share the common
feature of selecting value-objects from functionings and capabilities. Further, the capability
approach can be used with different methods of determining relative weights and different
mechanism for actual evaluation. The approach, if seen as a theory of algorithmic evaluation,
would be clearly incomplete.» (Sen, 1993:48). Neither does the non-weighing strategy seem to
be a useful aggregative route: «The varying importance of different capabilities is as much a part
of the capability framework as the varying value of different commodities is a part of the real
income framework. Equal valuation of all constitutive elements is needed for neither. We cannot
criticize the commodity-centered evaluation on the ground that different commodities are
weighted differently. Exactly the same applies to functionings and capabilities.» (Sen, 1992: 45-
46). In empirical terms, in CFM we decided the relative importance of each functioning on
objective grounds14, using a data-driven method independent of value judgments. More
specifically, we follow the path suggested by Chiappero Martinetti (1994: 383-384) and define
for each indicator of commodities determining functionings a weight wj based on the inverse
function of the frequency of the indicator itself in Italian regions:
Wj = log(1∕f )∕Σlog(1∕f ) (1)
with fi > 0 frequency of the i-th indicator under consideration15.
Therefore the essential character of the indicator is given by the diffusion it has in society: the
less it is widespread, the more it is relevant. Or the less the society has it, the more the society
values it. So, when an indicator shows a higher frequency of low values, the weight attached to
it will be greater then the one attached to another indicator showing lower frequencies and vice
versa (see infra 2.2 for a detailed example).
This overview of the methodological issues to be considered in the empirical application
could give the impression that dealing with the somehow elusive and incomplete soul of Sen’s
approach involves an inescapable difficulty. But incompleteness, far from being a pretext for the
persistence of the utilitarian perspective, guarantees the flexibility needed to adapt the exercise
to the ever-changing context. Postponing to the next section the practical and application-
oriented questions raised by CFM, there seems to be no major weakness from a methodological
point of view in the process of dynamic operationalization of the capability approach. There is
13 We assume that a substitute of GDP is useless and misguiding. Reality is too complex to be subsumed by a single
number: «The passion of aggregation makes good sense in many contexts, but it can be futile or pointless in
others» (Sen, 1987(b):33).
14 The adoption of a weighting scheme reflects the system of values of the society under observation. The definition
of the weights by the decision-maker according to her own preferences could be another alternative. To be
uncontroversial both the options share the need for certain principles of distributive justice and equity, whose
consideration is beyond the reach of this work. We therefore look for acceptability on the less theoretical ground
of quantitative objectivity.
15 The choice of the logarithm is intended «not to attribute an excessive importance to the indicators showing a too
low frequency», as Chiappero Martinetti states (1994: n. 19, p. 384).
9