(Table B1). The nature of the model now changes appropriately. That is, the least cost set of
feasible projects are selected in order to achieve the adjusted amount of increases in the level of
indicators.
We now attend to computing δjr . There are a number of ways of finding these coefficients. One
usual way is to assume that the policy maker would provide these values. However, we can
propose a number of ways for computing these coefficients.
(i) One simple way is to reduce Mrj by a constant percentage (for example reducing all Mrj
by 10%).
(ii) Another way is to reduce Mrj for the jth indicator (for all provinces) by a certain ratio.
(iii) A more accurate way is to reduce the change in targets for each province with respect to
the overall position of the province on the development scale RHDI. That is, setting
δjr for the province with the lowest ranking equal to 1 and scaling δjr for other provinces
proportionately according to their relative RHDI. It must be pointed out that in this
method we are in effect adjusting δr rather than δjr . In other words the weights for all
indicators belonging to province r are the same.
(iv) An even more accurate method, deals with a set of weights which vary not only with
respect to different provinces but also with respect to different indicators. They may be
obtained from the detailed elements of the RHDI. These are all (Zij -Zoj) for province
i (as computed from equation 5). For indicator j for the province furthest away from the
ideal province put δjr =1 and assign values to other δjr proportionately.
In applying the last method sometimes Mrj are negative implying a reduction in the level of
indicators. This happens when the computed targets Tjr is less than the actual value for the
indicator concerned. In such cases, if the policy maker does not wish to reduce the level of the
indicator it is appropriate to put the relevant δjr =0.
Table 4 displays the computed adjusted targets for indicator 2, adult literacy, for all provinces
employing method (iv) explained above.
15