A capability approach for seismic risk analysis and management - Seismic Risk Analysis and Management of Civil Infrastructure Systems - page 257

he or she could achieve the functionings that his or her peer chooses to
achieve.
Adequate indicator(s) for a given capability must satisfy two criteria
(Gardoni and Murphy 2009). First, an indicator must be representative, or
such that it tracks the level of a given capability in practice. Appropriate
indicators for a given capability may be to some extent hazard-specifi c.
Second, it should be reasonable in the sense that there is an intuitively
plausible relationship between the indicator and the capability at issue. This
second criterion is important for risk communication and effective policy-
making for natural hazards. If there is no intuitively plausible relationship
between a given indicator and its corresponding capability, this makes it
more diffi cult to communicate the results from a risk analysis to the general
public and to generate trust in the results.
Current engineering practice offers a rich set of potential indicators. The
information engineers currently use to assess the consequences of disasters
for communities, such as the number of individuals killed or injured, seems
to provide plausible data about how a community was affected. The capabil-
ity approach urges us to consider what kind of information the data is
providing, specifi cally in terms of what it is telling us about how specifi cally
the lives and well-being of individuals are affected.
There is a second way that current engineering practice can inform a
capability approach to risk. Engineering predictive models as well as past
data from actual disasters can inform the prediction of the likely value of
indicators (Gardoni and Murphy 2009). Predictions of the value of indica-
tors must account for uncertainties surrounding the magnitude and time of
occurrence of a given earthquake (or hazard more generally) and what the
impact of a given earthquake on a specifi c indicator will be. Confi dence
intervals can be used to specify the probability that the actual value of an
indicator will be within a given range. The mean can be used to specify the
expected value of a given indicator. A probability distribution can provide
a picture of the likelihood of each potential outcome, and so a picture of
the differential impact that hazards of different magnitudes might have.
Finally, it is important to note that it may be necessary to change the
indicators that are used to assess a particular capability, depending on the
period of time a given risk analysis is considering when determining
the potential impact of a hazard (Gardoni and Murphy 2009). Consider the
capability of being sheltered. The number of individuals homeless as a
result of a hazard might serve as an indicator for the initial impact of the
hazard. However, that indicator might not provide an accurate picture of
this capability over time. In the long term, few members of a community
might be literally homeless. However, housing for victims of a hazard might
be largely temporary, subsidized by the government for a specifi c period of
time. The temporary nature of such housing would render the capability