Python toolkit for measuring quality of life under spatial frictions

(c) Gabriel M. Ahlfeldt, Fabian Bald, Duncan Roth, Tobias Seidel

Version 1.0.2, 2024-10

This repository contains the Python implementation of the ARBSQOL-toolkit.

About

This toolkit implements a numerical solution algorithm to invert a quality of life (QoL) from observed data in various programming languages. The QoL measure is based on Ahlfeldt, Bald, Roth, Seidel (2024): Measuring quality of life under spatial frictions. Unlike the traditional Rosen-Roback measure, this measure accounts for mobility frictions—generated by idiosyncratic tastes and local ties—and trade frictions—generated by trade costs and non-tradable services, thereby reducing non-classical measurement error.

Notice that quality of life is identified up to a constant. Therefore, the inverted QoL measures measure has a relative interpretation only. We normalize the QoL relative to the first observation in the data set. It is straightforward to rescale the QoL measure to any other location or any other value (such as the mean or median in the distribution of QoL across locations).

When using this toolkit in your work, please cite Gabriel M. Ahlfeldt, Fabian Bald, Duncan Roth, Tobias Seidel (forthcoming): Measuring quality of life under spatial frictions.

Installation

To install the Python package of the ABRSQOL-toolkit, run the following command in your python environment in your terminal.

pip install ABRSQOL

Alternatively you can also install it from within your python script:

import subprocess, sys
subprocess.check_call([sys.executable, "-m", "pip", "install", 'ABRSQOL'])

In case an error occurs at the installation...

with an erorr message like 'metadata-generation-failed', it is likely caused by incompatabile versions of setuptools and packaging. This can be fixed by upgrading setuptools and packaging to compatible versions:

pip install --upgrade setuptools>=74.1.1
pip install --upgrade packaging>=22.0

Or by downgrading setuptools:

pip install --upgrade setuptools==70.0.0

Usage

You may then load the package by running:

import ABRSQOL

Or if you prefer alternatively import the function and testdata explicitly:

from ABRSQOL import invert_quality_of_life, testdata

Examples

Example 1: load testdata, run QoL inversion with default parameters, store result as 'QoL' variable, view result

testdata = ABRSQOL.testdata
testdata['QoL'] = ABRSQOL.invert_quality_of_life(df=testdata)
testdata.head()

Example 2: load your data from csv, run inversion, save result as csv

my_dataframe = read.csv("path/to/your/csv_filename.csv")
my_dataframe['quality_of_life'] = ABRSQOL.invert_quality_of_life(
  # supply your dataset as a dataframe
  df=my_dataframe,
  # specify the corresponding variable name for your dataset
  w = 'wage',
  p_H = 'floor_space_price',
  P_t = 'tradable_goods_price',
  p_n = 'local_services_price',
  L = 'residence_pop',
  L_b = 'L_b',
  # freely adjust remaining parameters
  alpha = 0.7,
  beta = 0.5,
  gamma = 3,
  xi = 5.5,
  conv = 0.3,
  tolerance = 1e-11,
  maxiter = 50000
)
# Write output to target folder (just replace the path)
from pandas import write_csv
write_csv(my_dataframe, 'C:/FOLDER/my_data_with_qol.csv')

Example 3: Reference variables in your dataset by using the column index

my_dataframe['QoL'] = ABRSQOL.invert_quality_of_life(
  df=my_dataframe,
  w = 1,
  p_H = 3,
  P_t = 4,
  p_n = 2,
  L = 6,
  L_b = 5
)

Example 4: Having named the variables in your data according to the default parameters, you can omit specifying variable names

my_dataframe['QoL'] = ABRSQOL.invert_quality_of_life(
  df=my_dataframe,
  alpha = 0.7,
  beta = 0.5,
  gamma = 3,
  xi = 5.5,
  conv = 0.5
)

Ready-to-use script

If you are new to Python, you may find it useful to execute the Example.py (or Example.ipynb) script saved in this folder. It will install the package, load the testing data set, generate a quality-of-life index, and save it to your working directory. It should be straightforward to adapt the script to your data and preferred parameter values.

Acknowledgements

The Python version of the toolkit was developed by Max von Mylius based on the Stata and MATLAB functions.