eyayaw/housing-supply-elasticity-in-germany

Replication code for my paper `Geographic determinants and the price elasticity of housing supply in Germany`

Replication code for my paper:

Geographic Constraints and the Housing Supply Elasticity in Germany — Ruhr Economic Papers #1003

Important

You can find the latest version in this repo Beze_2023_the-Price-Elasticity-of-Housing-Supply-in-Germany.pdf, or view it here.

Warning

RWI-GEO-RED data is not included in this repository. Thus, you need access to the data to reproduce this project.

Requirements

• R 4.3.3

• R packages:

  • data.table devel 1.15.99
  • more packages listed in renv.lock file

  This project depends heavily on the awesome data.table package.

  pkg	num_files
  rmarkdown	31
  data.table	30
  knitr	13
  sf	10
  readxl	10
  tools	9
  bookdown	5
  tibble	5
  ivreg	4
  ggplot2	4

• quarto to render the manuscript

Usage

1. Data

   1. Housing price: Accessed the data from the RWI FDZ at Ruhr

   2. Geodata:

      • administrative boundaries from the Federal Agency for Cartography and Geodesy, destatis.de and inkar.de

      • land cover data from the Copernicus Land Monitoring Service

      • Digital Terrain Model of Germany (DGM200) from BKG

   3. regional data from Regionaldatenbank Deutschland

2. R scripts: The order in which the scripts should be run is provided in main.R.

3. The index.qmd file generates the manuscript.

   • Render it with Cmd+Shift+K in RStudio/VS Code. Or, run the following code in the terminal:

     quarto render index.qmd
     # move figure notes out of the caption
     python3 tidy_figure_notes.py index.tex Beze_2023_the-Price-Elasticity-of-Housing-Supply-in-Germany.pdf

The price elasticity of housing supply estimates

The main results are provided in Table 2: IV Results: Housing Supply Elasticity Estimates of the manuscript.

If you want to calculate elasticity estimates for each county (kreis), compute “developed fraction” and “undevelopable fraction” at the county level. Then, use the coefficients estimated from the regression output for each variable to obtain estimates at the county level. It is important to note that the variation across counties is solely attributable to these two variables.

$$ \begin{aligned} &\Delta\ln H_i^s = \varepsilon^S\Delta\ln P_i + \beta^{\text{Developed}} \Delta \ln P_i\times\text{Developed}_i + \beta^{\text{Unavail}} \Delta \ln P_i\times\text{Unavail}_i + \boldsymbol{\beta}\mathbf{X}_i^S + u^S_i\\ &\implies \varepsilon^S_i = \varepsilon^S + \beta^{\text{Developed}}\text{Developed}_i + \beta^{\text{Unavail}}\text{Unavail}_i, \end{aligned} $$

where $i$ can be any spatial scale.

Please look at the script/constraints/compute_supply_constraints.R for constructing constraints at the county level.

---

Please email me if you find any mistakes or have any questions about the project.

Citation

BibTeX entry:

@book{beze_2023,
    address = {DE},
    title = {Geographic {Constraints} and the {Housing} {Supply} {Elasticity} in {Germany}},
    copyright = {All rights reserved},
    isbn = {978-3-96973-169-7},
    url = {https://doi.org/10.4419/96973169},
    publisher = {RWI},
    author = {Beze, Eyayaw},
    year = {2023},
}

For attribution of this repository, please cite the paper.