ABSTRACT
The tabular comparisons undertaken in Chapter 6 provide useful information concerning the 'yes/no' question: was the influence of Structural Adjustment Loans positive or not? However, this approach cannot tell us the precise quantitative influence of programme lending in relation to other influences; nor can it explicitly isolate the influence of exogenous non-policy variables, such as that of the weather and terms of trade movements, from the influence of programme lending; nor, within the category of programme lending, can it isolate the influence of Bank SALs from other Bank programme lending and Fund programmes, or the influence of policy changes attached to SALs from the influence of the money which they provide. To answer such questions we need a technique which will tell us what proportion of the variance in the Bank's chosen target variables is accounted for by specific independent variables. The appropriate technique for this purpose is multiple regression, and in this section we apply these methods to the same sample of SAL countries examined in Chapter 6, but now incorporating the influence of all policy-based lending rather than SALs alone. The five relationships which we wish to estimate each express the major targets of World Bank policy-based lending as a function of three groups of variables: financial flows from the IMF and World Bank, plus domestically financed investment; degree of compliance with Bank policy conditionality, as analysed in Chapter 5; and extraneous (i.e. non-aid) variables believed to have an influence on growth and the Bank's other policy targets. In its simplest terms this can be seen as a variant of the fundamental Hartod-Domar growth relationship: the growth rate equals savings, times the productivity of capital. The financial flow variables represent different components of savings; the level of compliance with conditionality and the extraneous variables represent obvious potential influences on the productivity of capital. In symbolic form the relationships we wish to estimate are:
