PROVIDE Project Technical Paper 2005:1
February 2005
90% 63.31995 3090.994 Variance 12751.18
95% 116.2348 5134.359 Skewness 26.65743
99% 336.1646 8509.819 Kurtosis 1499.697
Figure 11: Distribution of the relative income and expenditure difference (variable diffp)
Note: Only values between -100% and 100 included in the graph. The vertical lines represent (from left to right)
the 10th, 25th, 50th, 75th and 90th percentiles of variable diffp.
The fact that income or expenditure is underreported is not necessarily the problem, as this
is natural for most surveys of this kind. More problematic is the large average differences
between the two. A simple experiment performed here ranks households first by expenditure
(deciles) and then by income (deciles). Table 10 tabulates household income deciles against
household expenditure deciles. If income and expenditure were exactly the same, or even
within reasonable distances from each other, one would expect all households to lie on the
diagonal of the matrix. The shaded band above and below the diagonal shows those
households that move one group up or down. On average between 49.2% and 83.9% of
households remain in the same deciles. If the bands above and below are included, the figures
rise to between 81.6% and 94.8%.41
41 These percentages are reported in the last two rows.
55
© PROVIDE Project