Working Hours, GDP, and Productivity

You would think that the more hours you work, the more productive you are and the more money you make. However, some speculate that there is a balance with these factors because overworking diminishes motivation and productivity. There has been a recent push for a 4-day workweek, as opposed to the traditional 5-day workweek, in order to boost productivity and wealth. This project aims to study these relationships.

The Four-Day Workweek Merits Consideration - Forbes

The evidence overwhelmingly says 4-day work weeks are good for everyone

import pandas as pd
import matplotlib.pyplot as plt
import numpy as np

# import datasets
df_hours_and_GDP = pd.read_csv('annual-working-hours-vs-gdp-per-capita-pwt.csv')
df_productivity_and_hours = pd.read_csv('productivity-vs-annual-hours-worked.csv')

# merge dataframes
df_merged = pd.merge(df_hours_and_GDP, df_productivity_and_hours, how='left', left_on=['Entity', 'Year'], right_on = ['Entity', 'Year'])

# drop columns
df_merged_clean = df_merged.drop(columns=['Code_x', 'Code_y', 'Annual working hours per worker_y', 'Population (historical estimates)_y', 'Continent_y', 'Population (historical estimates)_x'])

# rename columns
df_merged_clean_renamed = df_merged_clean.rename(columns={'Entity': 'Country', 
            'Annual working hours per worker_x': 'Annual Working Hours per Worker',
            'GDP per capita (output, multiple price benchmarks)':'GDP per Capita',
            'Productivity: output per hour worked':'Productivity',
            'Continent_x':'Continent'})

# drop NaN (from certain columns)
df_all = df_merged_clean_renamed.dropna(subset=['Country', 'Year', 'Annual Working Hours per Worker', 'GDP per Capita', 'Productivity'])

# reset indexes
df2 = df_all.reset_index()
df_all = df2.drop(columns=['index'])

# present data
df_all

	Country	Year	Annual Working Hours per Worker	GDP per Capita	Continent	Productivity
0	Argentina	1950	2034.0000	2931.7388	NaN	3.727674
1	Argentina	1951	2037.8667	2940.7954	NaN	3.752668
2	Argentina	1952	2041.7408	2629.9502	NaN	3.365233
3	Argentina	1953	2045.6223	2747.4377	NaN	3.522695
4	Argentina	1954	2049.5112	2821.9634	NaN	3.623416
...	...	...	...	...	...	...
3487	Vietnam	2015	2191.3704	6180.3580	Asia	4.946606
3488	Vietnam	2016	2169.5515	6368.6510	NaN	5.156925
3489	Vietnam	2017	2131.9683	6841.6543	NaN	5.652919
3490	Vietnam	2018	2131.9683	7217.9240	NaN	5.982665
3491	Vietnam	2019	2131.9683	7506.8170	NaN	6.739149

3492 rows × 6 columns

x = df_all['Annual Working Hours per Worker']
y = df_all['GDP per Capita']

# plot GDP data
plt.scatter(x, y, c='#DE6666', s=1, alpha=0.6)
plt.xlabel('Annual Working Hours per Worker')
plt.ylabel('GDP per Capita')
plt.title('GDP per Capita vs. Annual Working Hours per Worker Across the World (1950 - 2019)')

#find line of best fit
model = np.poly1d(np.polyfit(x, y, 3))
polyline = np.linspace(1250, 3000)
plt.plot(polyline, model(polyline), color='#853e3e')

[<matplotlib.lines.Line2D at 0x124db1010>]

# plot productivity data
y = df_all['Productivity']
plt.scatter(x, y, c='#4772D1', label='Productivity', s=1, alpha=0.6)
plt.xlabel('Annual Working Hours per Worker')
plt.ylabel('Productivity')
plt.title('Productivity vs. Annual Working Hours per Worker Across the World (1950 - 2019)')

#find line of best fit
model = np.poly1d(np.polyfit(x, y, 3))
polyline = np.linspace(1250, 3000)
plt.plot(polyline, model(polyline), color='#12398c')

[<matplotlib.lines.Line2D at 0x124cea4d0>]

Data Dictionary

Country: country in which the data was sourced

Year: year in which the data was sourced

Annual Working Hours per Worker: average working hours per worker over an entire year

GDP per Capita: GDP per person in country, djusted for differences in the cost of living between countries, and for inflation

Continent: continent in which the data was sourced

Productivity: output per hour worked, measured as GDP per hour of work

Based on the size of the dataset and the relativley strong correlations shown, this data is sufficient to make progress on the problem at hand.

This project focuses on the correlations between GDP, annual hours worked, and productivity across the world in order to find the best balance between time at work and lesiure time to maximize productivity and therefore GDP. In future analyses, we will isolate the data by country and/or continent and find countries with the greatest GDP and lowest time spent at work. We will use this information to further investigate how these regions structure their workforce in order to apply their methods to other countries, specifically those who have very low GDP compared to the number of hours worked.