The following figure shows a scatter plot of household expenditures in Benin for 1996 and 1999.

1. The difference in the observed inequality between the two periods can be decomposed into vertical and horizontal inequality

• y₀                 : Vector of incomes of the initial period

• y₁                 : Vector of incomes of the final period

• E(y₁|y₀)     : Vector of the expected incomes in the final period

One can then decompose the observed variation in poverty into vertical and horizontal components of inequality such that:

• V component is the variation in poverty explained by vertical inequality

• H component is the variation in poverty explained by horizontal inequality

To compute all these components, we can use DAD to:

1. Generate the vector E(y₁|y₀). For this:
•  use the application : Distribution=> non parametric regression
• choose the appropriate vectors and parameters and click on the button GENERATE.

2. Once that this vector is generated, one can use the application Poverty=FGT Index (with 2 distributions) to finalise the computation.