ECNM10115 Programming and Numerical Methods for Economics

Lab 2

File from Learn - Course Material - Week 3

  • functions.py
  • UNPS_1112_PS2.xls

Import Libraries, Functions and Display Settings


import pandas as pd
# import numpy as np
# import matplotlib.pyplot as plt
import os
# import seaborn as sns

os.chdir('C:/Users/jzurita/OneDrive - University of Edinburgh/Courses/Programming Numerical Methods/Week 3/')

from functions import  gini

# import warnings
# warnings.filterwarnings('ignore')  #sns warning disactivated

#display options: To print entire tables
# pd.options.display.float_format = '{:,.2f}'.format
# pd.set_option('display.max_columns', None)  # or 1000
# pd.set_option('display.max_rows', None)  # or 1000
# pd.set_option('display.max_colwidth', None)  # or 199

# plots display default options
# sns.set_palette("Set2")   # set of colors in seaborn
# import matplotlib as mpl
# fm = mpl.font_manager

Import Data

data = pd.read_excel('UNPS_1112_PS2.xls')
data
hhid wave year region district county urban year_surv month_surv head_gender ... income wage_labor business_inc other_inc agriculture_inc livestock_inc wealth asset_value wealth_agrls land_value_hat
0 1013000204 2011-2012 2012 1 KALANGALA KYAMUSWA 0 2012 3 1.0 ... 717.277761 NaN 717.277761 0.000000 NaN NaN 102.530580 102.530580 NaN NaN
1 1013000206 2011-2012 2012 1 KAMPALA RUBAGA DIVISION 0 2012 9 1.0 ... 1839.529498 949.476802 890.052696 0.000000 NaN NaN 821.989843 821.989843 NaN NaN
2 1013000210 2011-2012 2012 1 KALANGALA KYAMUSWA 0 2012 2 1.0 ... 91.623072 NaN NaN 0.000000 NaN 91.623072 1202.661890 268.324710 934.337180 NaN
3 1013000212 2011-2012 2012 4 KYEGEGWA KYAKA 0 2012 3 1.0 ... 2617.802047 1989.529555 628.272491 0.000000 NaN NaN 494.764587 494.764587 NaN NaN
4 101300021302 2011-2012 2012 1 MPIGI MAWOKOTA 0 2012 4 2.0 ... 811.518634 NaN 157.068123 654.450512 NaN NaN 54.537543 54.537543 NaN NaN
... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ... ...
2608 4193003504 2011-2012 2012 4 KIRUHURA NYABUSHOZI 0 2012 2 2.0 ... 1046.796900 NaN NaN 0.000000 1037.634593 9.162307 11837.852383 256.108300 119.109993 11462.634090
2609 4193003506 2011-2012 2012 4 KIRUHURA NYABUSHOZI 0 2012 2 1.0 ... 10505.763174 NaN 8376.966549 130.890102 668.412123 1329.494399 6291.311323 405.759317 4794.940749 1090.611257
2610 4193003507 2011-2012 2012 4 KIRUHURA NYABUSHOZI 0 2012 2 2.0 ... 1216.754391 NaN NaN 0.000000 848.167863 368.586528 1888.801009 241.274089 271.596962 1375.929958
2611 4193003508 2011-2012 2012 4 KIRUHURA NYABUSHOZI 0 2012 8 1.0 ... 814.485477 NaN NaN 0.000000 814.485477 NaN 4142.348323 842.059658 10.471208 3289.817457
2612 4193003509 2011-2012 2012 4 KIRUHURA NYABUSHOZI 0 2012 8 1.0 ... 1937.435295 NaN NaN 0.000000 1314.136627 623.298667 21088.468461 796.248123 3429.320681 16862.899657

2613 rows × 30 columns

Note

Make sure you have installed xlrd package. Otherwise, you might have some difficulties to read the data file.

To install this package run:

conda install anaconda::xlrd

https://anaconda.org/anaconda/xlrd

!pip install xlrd 
Requirement already satisfied: xlrd in c:\users\jzurita\appdata\local\anaconda3\lib\site-packages (2.0.1)

Exercise 1

a). Are there duplicate households in the data? i.e. repeated household id?. How many observations?

data[data.duplicated(keep=False)]  
print('1a =======)')
print('There are no duplicate households')

# total number of observations
print('The total number of obsertations is N=', sum(data['hhid'].notna()))
1a =======)
There are no duplicate households
The total number of obsertations is N= 2613
  1. Present some basic summary statistics for some main variables. Anything we should be careful? like missing a lot of observations?
print('1b =======)')
print(data[['head_gender','head_age','familysize','consumption','income','wealth']].describe())
1b =======)
       head_gender     head_age   familysize  consumption        income  \
count  2597.000000  2597.000000  2597.000000  2613.000000   2613.000000   
mean      1.314209    46.068156     7.480554  1803.792687   1860.075795   
std       0.464289    15.068960     3.712526  1308.742941   2607.517603   
min       1.000000    14.000000     1.000000   250.305506     27.486921   
25%       1.000000    34.000000     5.000000   918.324958    471.204368   
50%       1.000000    44.000000     7.000000  1426.614855   1061.194208   
75%       2.000000    56.000000     9.000000  2296.859516   2234.800590   
max       2.000000   100.000000    33.000000  8369.898484  52137.736864   

             wealth  
count   2613.000000  
mean    4912.698146  
std     8359.886497  
min        0.000000  
25%      720.019916  
50%     2108.598380  
75%     5221.939857  
max    76396.339917

Exercise 2

a) Create the variables ‘log_c’ ‘log_inc’, ‘log_w’ that are the log of consumption, income, and wealth.

data[['consumption','income','wealth']].describe()  # 2613 observations. No observations below 0, but some equal to zero.
consumption income wealth
count 2613.000000 2613.000000 2613.000000
mean 1803.792687 1860.075795 4912.698146
std 1308.742941 2607.517603 8359.886497
min 250.305506 27.486921 0.000000
25% 918.324958 471.204368 720.019916
50% 1426.614855 1061.194208 2108.598380
75% 2296.859516 2234.800590 5221.939857
max 8369.898484 52137.736864 76396.339917 
# Create the logs
data[['log_c','log_inc','log_w']] = np.log(data[['consumption','income','wealth']])
data[['log_c','log_inc','log_w']] .describe()
log_c log_inc log_w
count 2613.000000 2613.000000 2613.000000
mean 7.273602 6.912703 -inf
std 0.670492 1.157038 NaN
min 5.522682 3.313710 -inf
25% 6.822551 6.155292 6.579279
50% 7.263060 6.967150 7.653779
75% 7.739298 7.711907 8.560624
max 9.032397 10.861644 11.243690 
##
data[['log_c','log_inc','log_w']] = data[['log_c','log_inc','log_w']].replace([-np.infty], np.nan)
### The log-variance measure of inequality is not the best when there might be a lot of zeros in the data.

data[['log_c','log_inc','log_w']].describe()
log_c log_inc log_w
count 2613.000000 2613.000000 2597.000000
mean 7.273602 6.912703 7.475775
std 0.670492 1.157038 1.636211
min 5.522682 3.313710 0.269188
25% 6.822551 6.155292 6.600676
50% 7.263060 6.967150 7.664869
75% 7.739298 7.711907 8.565685
max 9.032397 10.861644 11.243690 
# plot in the same graph the log_c and the log_inc. 
fig, ax = plt.subplots()
sns.distplot(data[['log_c']], label='Consumption')
sns.distplot(data[['log_inc']], label='Income')
plt.title('Consumption and Income Distributions: Uganda 2011-12')
plt.xlabel('Logarithms ')
plt.ylabel('Density')
ax.legend()
ax.grid()
plt.show()

e) Inequality compute the Gini of Consumption, Income and Wealth.

gini_ciw = data[['consumption','income','wealth']].apply(gini)
gini_ciw = pd.DataFrame(gini_ciw,columns=['Gini'])
print('    ')
print('2e =======')
print('Inequality in Uganda: Gini coefficients')
print(gini_ciw)
    
2e =======
Inequality in Uganda: Gini coefficients
                 Gini
consumption  0.365717
income       0.555572
wealth       0.656828

f) Compute the share of wealth of the bottom 50%.


shares = [0.5]
bottom = [0]
bottom.extend(shares) #[0 0.5]
    
var_array = np.sort(np.array(data['wealth'].dropna()))
var_total = sum(var_array) # Total Sum of Data Wealth
    
n= len(var_array)
percentiles = n*np.array(bottom)  # [0.0  0.5]
percentiles = percentiles.tolist()
percentiles = [int(x) for x in percentiles] ### Round to interger
    
mg_bottom= []
for i in range(1,len(percentiles)):  # len(percentiles)=2
    print(i)
    a = sum(var_array[percentiles[0]:percentiles[i]])/var_total
    mg_bottom.append(a)
    
              
mg_bottom
1
[0.08210562223168812]
mg_bottom
[0.08210562223168812]
percentiles
[0, 1306]
# f.  Compute the share of wealth of the bottom 50%.  that the top 10, 5, 1% in wealth hold. 
def bottom_shares(variable, dataset, shares = [0.5]):
    bottom = [0]
    bottom.extend(shares)
    
    var_array = np.sort(np.array(dataset[variable].dropna()))
    var_total = sum(var_array)
    
    n= len(var_array)
    percentiles = n*np.array(bottom)
    percentiles = percentiles.tolist()
    percentiles = [int(x) for x in percentiles] 
    
    mg_bottom= []
    for i in range(1,len(percentiles)):
        print(i)
        a = sum(var_array[percentiles[0]:percentiles[i]])/var_total
        mg_bottom.append(a)
              
    return mg_bottom

w_bottom = bottom_shares('wealth',dataset=data,shares=[0.25])
print('    ')
print('2f =======')
print('The Bottom 50% hold only 8% of the total wealth in Uganda')


def top_shares(variable, dataset, shares =[ 0.90, 0.95, 0.99]):

    shares.append(1)
    var_array = np.sort(np.array(dataset[variable].dropna()))
    var_total = sum(var_array)
    n= len(var_array)
    percentiles = n*np.array(shares) # multiple the vector by weights or shares. 
    percentiles = percentiles.tolist()
    percentiles = [int(x) for x in percentiles]
    
    mg_top = []
    for i in range(0,len(percentiles)-1):
        a = sum(var_array[percentiles[i]:percentiles[-1]])/var_total
        mg_top.append(a)               
    return  mg_top

w_top = top_shares('wealth',dataset=data,shares=[0.90, 0.95, 0.99])
print('the top 10%, 5% and 1% share 50%, 35%, and 11% of the wealth in the economy.')
1
    
2f =======
The Bottom 50% hold only 8% of the total wealth in Uganda
the top 10%, 5% and 1% share 50%, 35%, and 11% of the wealth in the economy.
(2613,)
percentiles
[0, 1306]
shares =[ 0.90, 0.95, 0.99]
shares.append(1)
shares
[0.9, 0.95, 0.99, 1]