This preview shows page 1. Sign up to view the full content.
Unformatted text preview: ary, you’re always gonna miss out, also when there are shifts within the same bracket of percentage) Mean absolute deviation: Absolute value (Sum every (Yi Ymean))/number of data points measure how far stretched out the distribution is applies anonymity, population, relative income, but fails Dalton Coefficient of variation Gini Coefficient To get Gini coefficient, you need to rank everybody, go line by line and say at every line WHAT FRACTION OF THE POPULATION YOU’VE COVERED and then another line that says WHAT PERCENTAGE OF THE INCOME DO THESE PEOPLE OWN ! plot those two as graph on excel Join tops of these histograms and you get LORENZ Curve Gini Coefficient is shaded area between line of equality and tops of Lorenz curve/ Total Area BCD (triangle formed if everyone had 0% but one person had 100%) Measures how bowed out the Lorenz curve is from line of equality Lorenz curve can never surpass line of equality (if you rank people should be able to see it goes from small number to big number) If Lorenz curves cross, then you can’t say anything about inequality HOW DO I KNOW WHICH ONE IS BETTER Some general principles that are used in constructing/choosing inequality measures: Distribution is set of (y1, y2, y3 … yN) with each y as an income group Anonymity principle Income distribution should not be influenced by the names or incomes behind the groups of people at either end of the spectrum, or certain peop...
View
Full
Document
 Fall '13
 RajDesai

Click to edit the document details