What you appear to have done is relabel your income variable categories with the income value to which the category refers. But this is not entirely clear: are you labelling the category with the income corresponding to the lower bound of the interval? (Or the upper bound?) Remember that an income category is defined by both a lower income value and an upper income value, and there are additional subtleties such as whether the category is defined as (1) greater than or equal to the lower bound value and less than the upper bound value, or (2) greater than the lower bound value and less than or equal to the upper bound value.
Second complication: your category boundaries are defined in terms of observed money income. However, I think you wish to define 'poverty' status in terms of a household's 'equivalent' (size-adjusted) income, where equivalent income is income divided by the square root of household size.
But what you know is whether a household's money income value falls within a particular money income category, not whether the household's size-adjusted income lies within some size-adjusted income category.
For a household of size 1, money income = size-adjusted income. However for a household of size S that has a money income lying in a particular category with bounds a and b, the corresponding size-adusted bounds are a/sqrt(S) and b/sqrt(S). You have different size-adjusted category information for households of different sizes.
Your poverty line definition is confusing to me. You refer to "1250" as the poverty line -- but does this refer to money income or per capita income or size-adjusted income, or what? Actually, from your later code, it appears that you wish to define the poverty line as a fraction of the median of size-adjusted income (not observed money income). One cannot do that very easily without first looking at the numbers of households falling with different size-adjusted income categories.
Bottom line: you have grouped data on money income, and so it is going to quite difficult to estimate where a household lies in the distribution of size-adjusted income.
NB In future, please use CODE delimiters to report the Stata commands you typed and output it generated. Please read the Forum FAQ about this.
Second complication: your category boundaries are defined in terms of observed money income. However, I think you wish to define 'poverty' status in terms of a household's 'equivalent' (size-adjusted) income, where equivalent income is income divided by the square root of household size.
But what you know is whether a household's money income value falls within a particular money income category, not whether the household's size-adjusted income lies within some size-adjusted income category.
For a household of size 1, money income = size-adjusted income. However for a household of size S that has a money income lying in a particular category with bounds a and b, the corresponding size-adusted bounds are a/sqrt(S) and b/sqrt(S). You have different size-adjusted category information for households of different sizes.
Your poverty line definition is confusing to me. You refer to "1250" as the poverty line -- but does this refer to money income or per capita income or size-adjusted income, or what? Actually, from your later code, it appears that you wish to define the poverty line as a fraction of the median of size-adjusted income (not observed money income). One cannot do that very easily without first looking at the numbers of households falling with different size-adjusted income categories.
Bottom line: you have grouped data on money income, and so it is going to quite difficult to estimate where a household lies in the distribution of size-adjusted income.
NB In future, please use CODE delimiters to report the Stata commands you typed and output it generated. Please read the Forum FAQ about this.
Comment