Global structure economy: A vision of GDP per capita and population

Explores the global distribution of total GDP through a multidimensional perspective, offering a comprehensive view of income distribution across the worldwide population.

Keywords

value chains

Summary

This animation illustrates how the hegemony exerted by certain nations has aimed to preserve an unequal income distribution by capturing the greater share of the value chains over the past decades.

# Libraries
# =====================================================================
import requests
import wbgapi as wb
import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt
import matplotlib.animation as animation
import matplotlib.ticker as ticker

# Data Extraction (Countries)
# =====================================================================
# Extract JSON and bring data to a dataframe
url = 'https://raw.githubusercontent.com/guillemmaya92/world_map/main/Dim_Country.json'
response = requests.get(url)
data = response.json()
df = pd.DataFrame(data)
df = pd.DataFrame.from_dict(data, orient='index').reset_index()
df_countries = df.rename(columns={'index': 'ISO3'})

# Data Extraction - WBD (1960-1980)
# ========================================================
# To use the built-in plotting method
indicator = ['NY.GDP.PCAP.CD', 'SP.POP.TOTL']
countries = df_countries['ISO3'].tolist()
data_range = range(1960, 2024)
data = wb.data.DataFrame(indicator, countries, data_range, numericTimeKeys=True, labels=False, columns='series').reset_index()
df_wb = data.rename(columns={
    'economy': 'ISO3',
    'time': 'Year',
    'SP.POP.TOTL': 'LP',
    'NY.GDP.PCAP.CD': 'NGDPDPC'
})

# Adjust LP and filter before 1980
df_wb['LP'] = df_wb['LP'] / 1000000
df_wb = df_wb[df_wb['Year'] < 1980]

# Data Extraction - IMF (1980-2030)
# =====================================================================
#Parametro
parameters = ['LP', 'NGDPDPC']

# Create an empty list
records = []

# Iterar sobre cada parámetro
for parameter in parameters:
    # Request URL
    url = f"https://www.imf.org/external/datamapper/api/v1/{parameter}"
    response = requests.get(url)
    data = response.json()
    values = data.get('values', {})

    # Iterate over each country and year
    for country, years in values.get(parameter, {}).items():
        for year, value in years.items():
            records.append({
                'Parameter': parameter,
                'ISO3': country,
                'Year': int(year),
                'Value': float(value)
            })
    
# Create dataframe
df_imf = pd.DataFrame(records)

# Pivot Parameter to columns and filter nulls
df_imf = df_imf.pivot(index=['ISO3', 'Year'], columns='Parameter', values='Value').reset_index()

# Filter after 2024
df_imf = df_imf[df_imf['Year'] >= 1980]

# Data Manipulation
# =====================================================================
# Concat and filter dataframes
df = pd.concat([df_wb, df_imf], ignore_index=True)
df = df.dropna(subset=['NGDPDPC', 'LP'], how='any')

# Merge queries
df = df.merge(df_countries, how='left', left_on='ISO3', right_on='ISO3')
df = df[['ISO3', 'Country', 'Year', 'LP', 'NGDPDPC', 'Analytical', 'Region']]
df = df[df['Region'].notna()]

# Create a list
dfs = []

# Interpolate monthly data
for iso3 in df['ISO3'].unique():
    temp_df = df[df['ISO3'] == iso3].copy()
    temp_df['Date'] = pd.to_datetime(temp_df['Year'], format='%Y')
    temp_df = temp_df[['Date', 'LP', 'NGDPDPC']]
    temp_df = temp_df.set_index('Date').resample('ME').mean().interpolate(method='linear').reset_index()
    temp_df['ISO3'] = iso3
    temp_df['Year'] = temp_df['Date'].dt.year 
    dfs.append(temp_df)
    
# Concat dataframes    
df = pd.concat(dfs, ignore_index=True)

# Filter nulls and order
df = df.sort_values(by=['Year', 'NGDPDPC'])

# Calculate 'left accrual widths'
df['LP_Cum'] = df.groupby('Date')['LP'].cumsum()
df['LP_Per'] = df['LP'] / df.groupby('Date')['LP'].transform('sum')
df['LP_Cum_Per'] = df['LP_Cum'] / df.groupby('Date')['LP_Cum'].transform('max')
df['Left'] = df['LP_Cum_Per'] - df['LP_Per']

# Calculate GDP Average weighted by Population and partitioned by Year
df['AVG_Weight'] = df.groupby('Date')['NGDPDPC'].transform(lambda x: np.average(x, weights=df.loc[x.index, 'LP']))

# Add a total GDP column and cummulative it
df['GDP'] = df['NGDPDPC'] * df['LP']
df['GDPcum'] = df.groupby('Date')['GDP'].cumsum()
df['GDP_Per'] = df['GDP'] / df.groupby('Date')['GDP'].transform('sum')
df['NGDPDPC_Change'] = ((df['NGDPDPC'] / df.groupby('ISO3')['NGDPDPC'].transform('first')) - 1) * 100

# Copy a df sample to calculate a median
df_sample = df.copy()
columns = df.columns
df_sample = np.repeat(df_sample.values, df_sample['LP'].astype(int), axis=0)
df_sample = pd.DataFrame(df_sample, columns=columns)
df_sample.dropna(subset=['NGDPDPC'], inplace=True)
df_sample['NGDPDPC'] = pd.to_numeric(df_sample['NGDPDPC'], errors='coerce')
df_sample['GDPcum'] = df_sample.groupby('Date')['NGDPDPC'].transform('cumsum')

# Function to get median row
def medianrow(group):
    medianrow = len(group) // 2  # Índice de la mitad
    return group.iloc[medianrow]
df_sample = df_sample.groupby('Date').apply(medianrow).reset_index(drop=True)

# Transform columns
df_sample = df_sample[['Date', 'NGDPDPC', 'GDPcum']].reset_index(drop=True)
df_sample = df_sample.rename(columns={'NGDPDPC': 'Median', 'GDPcum': 'Median_Total'})

# Merge queries
df = df.merge(df_sample, how='left', on='Date')

# Add percentage bottom 50
df['Median_Total_Per'] = df['Median_Total'] / df.groupby('Date')['GDP'].transform('sum')

print(df)
 
# Data Visualization
# =====================================================================
# Seaborn figure style
sns.set(style="whitegrid")
fig, ax = plt.subplots(figsize=(16, 9))

# Create a palette
palette = sns.color_palette("coolwarm", as_cmap=True).reversed()

# Function to refresh animation
def update(date):
    plt.clf()
    subset = df[df['Date'] == date]
    subset_usa = subset[subset['ISO3'] == 'USA'].copy()
    
    # Normalize GDPcum in a range [0, 1]
    gdp_min = subset['GDPcum'].min()
    gdp_max = subset_usa['GDPcum'].max()
    norm = plt.Normalize(gdp_min, gdp_max)
    colors = palette(norm(subset['GDPcum']))
    
    # Create a Matplotlib plot
    bars = plt.bar(subset['Left'], subset['NGDPDPC'], width=subset['LP_Per'], 
            color=colors, alpha=1, align='edge', edgecolor='grey', linewidth=0.1)
    
    # Configuration grid and labels
    plt.text(0, 1.05, 'Distribution of Global GDP', fontsize=13, fontweight='bold', ha='left', transform=plt.gca().transAxes)
    plt.text(0, 1.02, 'Evolution of Countries from 1960 to 2029', fontsize=9, color='#262626', ha='left', transform=plt.gca().transAxes)
    plt.xlim(0, subset['LP_Cum_Per'].max())
    plt.ylim(0, subset_usa['NGDPDPC'].max() * 1.05)
    plt.grid(axis='x')
    plt.grid(axis='y', linestyle='--', linewidth=0.5, color='lightgray')
    plt.xlabel('Cumulative Global Population (%)', fontsize=10, fontweight='bold')
    plt.ylabel('GDP per capita (US$)', fontsize=10, fontweight='bold')
    plt.tick_params(axis='x', labelsize=9)
    plt.tick_params(axis='y', labelsize=9)
    plt.gca().xaxis.set_major_formatter(ticker.FuncFormatter(lambda x, _: f'{int(x*100):,}%'))
    plt.gca().yaxis.set_major_formatter(ticker.FuncFormatter(lambda x, _: f'{int(round(x, -3)):,}'))
    plt.xticks(np.linspace(0, subset['LP_Cum_Per'].max(), 5))
    plt.yticks(np.linspace(0, subset_usa['NGDPDPC'].max() * 1.05, 8))
        
    # Add Labels to relevant countries
    for bar, value, country in zip(bars, subset['GDP'], subset['ISO3']):
        if country in ['CHN', 'IND', 'USA', 'IDN', 'PAK', 'NGA', 'BRA', 'BGD', 'RUS', 'MEX', 'JPN', 'VNM', 'DEU', 'GBR']:
            plt.gca().text(bar.get_x() + bar.get_width()/2, bar.get_height(),
                        f'{country}\n{''}', ha='center', va='bottom', fontsize=7, color='grey')

    # Add Median Line and Label
    median = subset['Median'].max()
    median_total = subset.iloc[0]['Median_Total']
    median_total_per = subset.iloc[0]['Median_Total_Per']
    maxis = subset_usa['NGDPDPC'].max()
    
    plt.axhline(
        y=median,
        color='darkred', 
        linestyle='--', 
        linewidth=0.5)

    plt.text(
        x=subset['Left'].max() * 0.02,
        y=median + (maxis * 0.06),
        s=f'GDP Capita Median: {median:,.0f}',
        color='darkred',
        verticalalignment='bottom',
        horizontalalignment='left',
        fontsize=10,
        weight='bold') 

    plt.gca().text(
                subset['Left'].max() * 0.02,
                median + (maxis * 0.04),
                f'Bottom 50 - Total GDP: {median_total:,.0f} MM ({median_total_per * 100:.2f}%)', 
                ha='left', va='center', 
                fontsize=9, 
                color='#737373')
    
    # Add USA Line and Label
    pibc_usa = subset_usa.iloc[0]['NGDPDPC']
    pib_usa = subset_usa.iloc[0]['GDP']
    pib_usa_per = subset_usa.iloc[0]['GDP_Per']
    
    plt.axhline(
        y=pibc_usa, 
        color='darkblue', 
        linestyle='--', 
        linewidth=0.5)
    
    plt.text(
        x=subset['Left'].max() * 0.02,
        y=pibc_usa * 0.95,
        s=f'GDP Capita USA: {pibc_usa:,.0f}',
        color='darkblue',
        fontsize=10,
        verticalalignment='bottom',
        horizontalalignment='left',
        weight='bold')

    plt.gca().text(
                subset['Left'].max() * 0.02,
                pibc_usa * 0.93,
                f'USA - Total GDP: {pib_usa:,.0f} MM ({pib_usa_per * 100:.2f}%)', 
                ha='left', va='center', 
                fontsize=9, 
                color='#737373')

    # Cover countries going outside plot
    formatted_date = date.strftime('%Y-%m') 
    plt.text(1, 1.15, f'{formatted_date}',
             transform=plt.gca().transAxes,
             fontsize=22, ha='right', va='top',
             fontweight='bold', color='white',
             bbox=dict(facecolor='white', edgecolor='none', boxstyle='square,pad=1'))

    # Add Year label
    plt.text(1, 1.06, f'{formatted_date}',
             transform=plt.gca().transAxes,
             fontsize=22, ha='right', va='top',
             fontweight='bold', color='#D3D3D3',
             bbox=dict(facecolor='white', edgecolor='none', boxstyle='square,pad=0.3'))
    
    # Add Data Source
    plt.text(0, -0.1, 'Data Source: IMF World Economic Outlook Database, 2024 | World Bank national accounts data, and OECD National Accounts data files.', 
            transform=plt.gca().transAxes, 
            fontsize=8, 
            color='gray')
    
    # Add author
    plt.text(1, -0.1, '@guillemmaya.com', 
         transform=plt.gca().transAxes, 
         fontsize=8, 
         color='#212121', 
         ha='right')

    # Add label "poorest" and "richest"
    plt.text(0, -0.065, 'Poorest',
             transform=ax.transAxes,
             fontsize=12, fontweight='bold', color='darkred', ha='left', va='center')
    plt.text(0.95, -0.065, 'Richest',
             transform=ax.transAxes,
             fontsize=12, fontweight='bold', color='darkblue', va='center')

# Configurate animation
dates = sorted(df['Date'].unique())
ani = animation.FuncAnimation(fig, update, frames=dates, repeat=False, interval=500, blit=False)

# Save the animation :)
ani.save('C:/Users/guill/Downloads/FIG_GDP_Capita_Bars_1960.webp', writer='imagemagick', fps=15)

# Print it!
plt.show()