Changes in the mean absolute difference (relating to the Gini coefficient)

89 Views Asked by At

Consider the expression for one half of the mean absolute difference divided by the mean $\mu$, i.e., $\frac{1}{2\mu{n^2}}\Sigma^n_{i=1} \Sigma^n_{j=1} |y_i-y_j|$ (the Gini coefficient). How does it change if we add one more entry $y_{n+1}$ to the array $y_1$, $y_2$, ..., $y_n$ that we are considering? Can we say when it is going to increase, decrease or stay constant, depending on $y_{n+1}$? Or, in other words, can you solve the following for $y_{n+1}$?

$\frac{1}{2\mu{n^2}}$ $\Sigma^n_{i=1} \Sigma^n_{j=1} |y_i-y_j| = \frac{1}{2\mu^\prime{(n+1)^2}}$ $\Sigma^{n+1}_{i=1} \Sigma^{n+1}_{j=1} |y_i-y_j| $,

where $\mu=\frac{1}{n}\Sigma^n_{i=1}y_i$ and $\mu^\prime=\frac{1}{n+1}\Sigma^{n+1}_{i=1}y_i$.

Thanks.