Abstract:
This paper addresses the "dependent variable" problem in food security policies, namely
the difficulty in classifying food security policies that limits comparative policy studies.
Policy comparisons require criteria that are general enough for broad application but
sensitive to the context. A rigorous and objective basis for comparisons would allow for
studying how policies, and food security policies in particular, emerge. This is important
in Africa as, in the past, food security and nutrition crises have been attributed to the
failure of government policies. This paper reviews the main key available public policy
classifications based on their predictability, mutual-exclusivity and relevance. These
include Lowi's and Wilson's typologies, the agricultural policies' classification by Norton
and the FAO-FAPDA classification. The review found that available typologies do not
accommodate multi-sectoral actions and are not entirely applicable to food security
public policy classification. The domain shift from food policy to food security, and more
recently to food systems demands that all elements in the food system to be taken into
consideration in the policy process. This limits the use of policies as "dependent
variables" and hence the study of how they emerge, particularly in Africa. A critique of
available policy classes shows that these cannot be treated as "dependent variables". It is
argued that a potential solution to the "dependent variable" problem of food security
policies lies in the development of a taxonomy, simplifying their complexity with
analytical shortcuts. Having reviewed Candel and Daugbjergs’ recent taxonomy,
refinements are proposed to be applied in the African context. The proposed taxonomy
represents an alternative to classify food security policies in Africa along four core
dimensions. This classification offers prospects for researchers to study what factors
drives policy-classes in one direction or the other, along the four dimensions. Although
the scales and calibration of the four dimensions will need to be developed and tested,
the proposed typology offers a way to treat the dimensions as “dependent variables”.
