A new and easy-to-use measure of literacy, its axiomatic properties and an application (Q2272181)

The authors consider measures of illiteracy of the form \(f(n,r,i)\) which take into account the number \(n\) of adults in the population, the number \(r\) of literates and the number \(i\) of ``isolated illiterates'' (i.e. the illiterates who live amidst people who are all illiterate). Three simple axioms for such measures \(f\) are proposed: (N) Normalization: \(f(n,r,n)=0\), \(f(n,r,0)=r/n\). (M) Monotonicity: \(f(n,r+1,i)>f(n,r,i)\), \(f(n,r,i-1)>f(n,r,i)\). (D) Weak decomposability: if \(f(n_1,r_1,i_1)=f(n_2,r_2,i_2)=c\) then \(f(n_1+n_2,r_1+r_2,i_1+i_2)=c\). It is shown that \(f\) satisfies (N), (M), (D) and some additional axiom I (invariability) if and only if for some \(\alpha\) \(f(n,r,i)={(1-\alpha)r\over (1-\alpha)n+\alpha i}\). An application to the data on illiteracy in South Africa is considered.