Poverty headcount ratio

[Shiran Sandarathne]

Hi, Can anyone please tell me what is the STATA command to calculate poverty headcount ratio in STATA version 12?

Thanks

[Kerri Agnew]

You might have a variable that shows what each person lives on, I can't remember the exact code I used, but I just created a dummy variable to signify if the person is below the poverty or not (=0 if they earn above the poverty line and =1 below poverty line. Then I tabbed the value of that dummy variable.to get the HCR. There might be some inbuilt commands to use, but I am unsure about what they are.

[Richard Williams]

What are your variables? What is your unit of analysis, e.g. individuals, nations? Kerri's calculation sounds right to me (multiplied by 100 maybe?) but it depends on how your data are structured.

[Patrick Kilgallen]

This works: @ http://sites.lafayette.edu/stifeld/f...10/06/Lab3.pdf

scalar z = 736
gen pcexp = hhexpend / hhsize
gen ind = 0
replace ind = 1 if pcexp <= z
gen gap0 = ind * ((z-pcexp)/z)^0

z is the poverty line which you must specify yourself (you could use 60% of median income/expenditure-that's one commonly used measure)

ind is a dummy =0 if the observation is below the line.

gap 0 is the headcount ration where pcexp is household expenditure / household size necessary to equivalise the income.



gen gap1 = ind * ((z-pcexp)/z)^1
gen gap2 = ind * ((z-pcexp)/z)^2
sum gap0 gap1 gap2

This part gives you two more measures where the third is foster greer thorbecke measure. More detail in the link.

[Shiran Sandarathne]

Thank you all..

Richard, my unit of analysis is households . I want to get the percentage of households (of the sample) below the poverty line. My variable is Per capita expenditure. I think Just taking the number of BPL households as a percentage of total households doesn't give me the correct answer as I need to incorporate sample weights as well.

Yeah as you say Kerri's answer ,ultiplied by 100 gives me somewhat reasonable number as well though not incorporated sample weights..

[Richard Williams]

If you have already created the poverty dummy, then it could be as simple as

svy: proportion poverty

Example:

webuse nhanes2f, clear
proportion diabetes
svy: proportion diabetes

I am assuming that you have pweights and have (or could) svyset your data. If you have weights but haven't svyset, try

proportion diabetes [pw = finalwgt]

If this doesn't work for your situation, provide some more details as to what your data set is like.

[Shiran Sandarathne]

Hi Richard,

Thank you very much. I have only weights not svyset. The last code works!! Thanks

[Richard Williams]

You may want to svyset your data, and you definitely want to if you also have information on clustering and stratification in your data. See http://www.stata.com/manuals13/svy.pdf, especially the opening intro section.

[Sergiy Radyakin]

Hi, Can anyone please tell me what is the STATA command to calculate poverty headcount ratio in STATA version 12?

Thanks

See: http://www.worldbank.org/adept
ADePT is using Numerics By Stata.
Here is the manual
For poverty-related questions see this book which explains how to use the ADePT software for analysis.

Best, Sergiy Radyakin

[Stephen Jenkins]

Have a look at povdeco on SSC, and references therein.

[GabrielBonfiglio]

The United States statistics on poverty rate standing was released. A comparison of financial status information from the previous years and up to present shows that poverty rates changed and the median income declined. Article source: visit nowadays with some of our web site.

[KAYDEN MARLI]

i have a question i want to calculate poverty rates but my income variable is in classes from No income to the highest class which is ($ 3601 - $5300 ) what should i do if the poverty line is $1100

[Stephen Jenkins]

You can't derive a point estimate for the poverty rate (proportion with income below the poverty line) without making further assumptions about the nature of the income distribution within the income class that $1100 is located. However, you can straightforwardly derive bounds on the poverty rate. You can report the proportion of persons with an income below the income defining the class lower bound, and proportion of persons with an income below the income defining the class upper bound. Re derivation of point estimates: fitting an income distribution to grouped data is a tricky business but can be done. (I suggest reading around the literature on this.)

[KAYDEN MARLI]

thank u stephen, i m new just learning to use stata and its quite tricky, but surely i will have to go through the books to understand.

[KAYDEN MARLI]

i have relabeled the income variable in median income so its now a continu variable what are the commands if i want to calculate poverty line for a household with a child (which i have generated?





here i have assigned a weight for every child in the household

var q12 = the number of children in the household

egen kindhh = sum (q12)
gen eqs_kind =.
replace eqs_kind = 1 +(kindhh *0.3)
tab eqs_kind


**** here i have recode the income classes into median income
gen str_inkomenklm = ""
replace str_inkomenklm = "0" if (q10new==0)
replace str_inkomenklm = "200" if (q10new==1)
replace str_inkomenklm = "500" if (q10new==2)
replace str_inkomenklm = "700" if (q10new==3)
replace str_inkomenklm = "850" if (q10new==4)
replace str_inkomenklm = "950" if (q10new==5)
replace str_inkomenklm = "1100" if (q10new==6)
replace str_inkomenklm = "1250" if (q10new==7)
replace str_inkomenklm = "1400" if (q10new==8)
replace str_inkomenklm = "1650" if (q10new==9)
replace str_inkomenklm = "1900" if (q10new==10)
replace str_inkomenklm = "2150" if (q10new==11)
replace str_inkomenklm = "2450" if (q10new==12)
replace str_inkomenklm = "2850" if (q10new==13)
replace str_inkomenklm = "3350" if (q10new==14)
replace str_inkomenklm = "4450" if (q10new==15)
replace str_inkomenklm = "5300" if (q10new==16)



**************** IF the povertly line income in our country is 1250 , IS THIS CORRECT TO CALCULATE childpoverty RATE

gen miss_comp = 0
quietly drop if miss_comp==1

global topline = 10*r(p50)
*Apply top and bottom codes
replace loonimpute = 0 if loonimpute <1250
replace loonimpute = $topline if loonimpute >$topline
*Apply lis equivalence scale
gen loonimpute_tb = loonimpute/(q12^0.5)
* Defining the poverty line as 50% of the median equivalised income
gen byte poor=(loonimpute_tb<r(p50)*0.5)

* the proportion households in poverty
sum poor [iw=fcalweight]
* the proportion individuals in poverty
sum poor [iw=fcalweight*eqs_kind]




*/