requisites for later values of both instruments (growth) and objectives (poverty reduction).
At a more micro level, improving gender equality is likely to be a prerequisite for halting and
reversing the spread of HIV/AIDS (part of Goal 6). Thus, some variables may start as
objectives but later be instruments for the achievement of other objectives.
Measurement problems
We have seen how interdependencies complicate strategies for achieving the MDGs, here
conceptualised as a mapping from instruments to objectives. Trade-offs and
complementarities also mean that measuring progress towards the targets in terms of
movements in the values of objectives and instruments may necessitate more sophisticated
analytical frameworks than those currently used. Benchmarking methods in their simplest
but common form measure progress by extrapolating past rates of change for each target
individually to see if the country in question is “on course” (Devarajan et. al. (2002); Sahn
and Stifel (2001)). Clearly this is a highly imperfect approach if interdependencies and
sequencing requirements are present. In such situations, measurement needs to take into
account of interdependencies at the level of both objectives and instruments to give a more
comprehensive picture of countries’ distance from the targets.
Measurement is also hampered by imperfect information about the actual values of
instruments and variables (point (9)). This is especially so because the MDG targets
encompass a broad range of variables, including “soft” social indicators. In addition,
interdependencies mean that the data problems of certain targets and instruments may lead
to difficulties in the assessment of related variables. Imperfections in data sets essential for
measurement of MDG outcomes and instruments including those for income, inequality,
poverty and health status are well known. Even for developed economies, data sets such as
those for inequality need to be handled with caution (Atkinson and Brandolini (2001)); in
poor countries where statistical capacity is limited this is even more critical (Srinivasan
(1994)).
Valuing outcomes
“Progress” implies valuation; in multi-dimensional space this is not straightforward. If
interdependencies—particularly trade offs—along with the requirements of sequencing
mean that not all goals can be achieved simultaneously, then social weightings will be
needed to arbitrate between targets. This suggests the need for a social welfare function
defined across the targets with explicit weights assigned to each objective. How such
weightings would be determined is a complicated political as well as economic problem
outside of the scope of this paper.
However, if choices do need to be made it would be better to make them explicitly rather
than by ad hoc means or in reaction to uncontrollable events. Current methods of
assessment do not fully acknowledge this and fail to take a systematic social welfarist
approach to the problem. Defining a social welfare function is a complex task but opens up
solutions to many of the problems flowing from the MDGs’ formulation as a set of fixed
targets. This is discussed further in section seven.