Wealth-index is a measure of the cumulative living standrd of a houselhold. It is calculated using data on assets posessed by households such as vehicles, clean water, electricity etc. This document aims at assesing the living standard of kenyans accross various regions(provinces) by measuring their wealth Indices. Calculation of wealth index will take the following simple steps 1. Selecting the variables 2. Exploratory data analysis 3. Recoding the data into binary variables 4. Conduct Principle Component analysis 5. Create wealth index quintiles 6. Graph the index and report

The data regarding assets has been collected and stored in spss (.sav) format. It is precoded as follows

##       Provinces code
## 1       central    2
## 2         coast    3
## 3       eastern    4
## 4 north eastern    5
## 5        nyanza    6
## 6   rift valley    7
## 7       western    8
library(haven)
library(tidyverse)
library(dplyr)
assets <- read_sav("C:/Users/Christine/Desktop/Rpubs/ASSETS.sav")
vars_labelled = map(assets, function(x) attr(x, "class") == "haven_labelled") %>% 
  unlist() %>% 
  names()
assets_factor = assets %>%
  mutate_at( vars(vars_labelled), as_factor)
head(assets_factor,3 )
## # A tibble: 3 x 24
##   aprovinc bicycle Motobike Radio Telephone Refrigerator Fan   Buckets Bed  
##   <fct>    <fct>   <fct>    <fct> <fct>     <fct>        <fct> <fct>   <fct>
## 1 eastern  no      no       no    yes       no           no    yes     yes  
## 2 nyanza   no      no       no    yes       no           no    yes     no   
## 3 central  no      no       no    yes       no           yes   yes     yes  
## # ... with 15 more variables: Bedsheets <fct>, Blankets <fct>, Nets <fct>,
## #   tables <fct>, Ox <fct>, Plough <fct>, Hoes <fct>, Axes <fct>, Muo <fct>,
## #   Kosiowo <fct>, Ngomo <fct>, Bellows <fct>, Jembe <fct>, Panga <fct>,
## #   Other <fct>
##                
##                 yes  no
##   central        74 503
##   coast           0   0
##   eastern       227 944
##   north eastern   0   0
##   nyanza          3 505
##   rift valley    91 537
##   western       115 424
##                
##                 yes  no
##   central        11 418
##   coast           0   0
##   eastern        97 848
##   north eastern   0   0
##   nyanza         15 341
##   rift valley     4 367
##   western        11 342
##                
##                  yes   no
##   central          1  576
##   coast            0    0
##   eastern          6 1165
##   north eastern    0    0
##   nyanza           3  505
##   rift valley      0  628
##   western          2  537



The tables above show wealth distribution in terms of possession of selected goods in the household. Now we use the above information to calculate the actual living standard among kenyans. The data will be coded in binary, 1 to mean possession of the household item being measured and 0 otherwise . Missing values will be assumed as zero as shown below.

assets[,2:24][assets[,2:24]=="2"]<-0
vars_labelled = map(assets, function(x) attr(x, "class") == "haven_labelled") %>% 
  unlist() %>% 
  names()
print(vars_labelled)
##  [1] "aprovinc"     "bicycle"      "Motobike"     "Radio"        "Telephone"   
##  [6] "Refrigerator" "Fan"          "Buckets"      "Bed"          "Bedsheets"   
## [11] "Blankets"     "Nets"         "tables"       "Ox"           "Plough"      
## [16] "Hoes"         "Axes"         "Muo"          "Kosiowo"      "Ngomo"       
## [21] "Bellows"      "Jembe"        "Panga"        "Other"
assets.num = assets %>%
  mutate_at( vars(vars_labelled), as.integer)
head(assets.num, 3)
## # A tibble: 3 x 24
##   aprovinc bicycle Motobike Radio Telephone Refrigerator   Fan Buckets   Bed
##      <int>   <int>    <int> <int>     <int>        <int> <int>   <int> <int>
## 1        4       0        0     0         1            0     0       1     1
## 2        6       0        0     0         1            0     0       1     0
## 3        2       0        0     0         1            0     1       1     1
## # ... with 15 more variables: Bedsheets <int>, Blankets <int>, Nets <int>,
## #   tables <int>, Ox <int>, Plough <int>, Hoes <int>, Axes <int>, Muo <int>,
## #   Kosiowo <int>, Ngomo <int>, Bellows <int>, Jembe <int>, Panga <int>,
## #   Other <int>


Wealth index is caclulated using Principle component analysis. This technique will take into account the components that portray the highest variability within the data, while minimizing autocorrelation. The first component will be used to construct wealth quintiles, ranked from 1 to 5, with 1 being the lowest and 5 the highest. 1 will represent the poorest in the society and 5 the wealthiest.

library(psych)
assets.num[is.na(assets.num)]<-0
assets.num$aprovinc<-as.factor(assets.num$aprovinc)
assets.pca<-psych::principal(assets.num[,2:24], rotate="varimax", nfactors=3,covar=T, scores=TRUE)
wealth_index=assets.pca$scores[,1]
Assets.indexed<-mutate(assets_factor,quintile=as.factor(cut(wealth_index,breaks=5,labels= c(1,2,3,4,5))))



ggplot(Assets.indexed, aes(aprovinc)) + geom_bar(aes(fill = quintile), position = "fill")+ xlab("Province")+ylab("Percentage")+ggtitle("Wealth by Province")



From the wealth index graph, Eastern seems to be the poorest province, followed by rift valley. Central province and rift- valley are the richer provinces.