RDP 2021-07: Macroprudential Limits on Mortgage Products: The Australian Experience Read me

This ‘read me’ file describes the supplementary materials provided with RDP 2021-07. The RDP uses confidential and subscription-based data, and the supplementary materials have been modified to remove anything that cannot be shared. In each of these instances of modification the files contain the text ‘REMOVED’. This means readers will be unable to run the code in these files. Still, the code provides the full set of details for how the variables were constructed and the regressions specified.

The ‘Data pull and formatting’ folder contains files that read and format data to subsequently use in regressions run in Stata. The files include R scripts, and csv data files that the R scripts read and write. These files were run by placing them together in the same R Project. The R scripts are:

• Data_pull_and_combine.R: queries bank data from an internal RBA database (SAFFI); reads the SAFFI_itemcode_names.csv file to apply variable names to the raw SAFFI data; reads the CANSTAR_data.csv file containing the interest rate data; reads the credit_aggs_gr_from_rba_website.csv raw data file, which contains credit growth data from RBA Statistical Table D1; and produces the data_for_macropru_analysis.csv data file, which is the data source used for most of the regressions in Stata.
• LVR_dataset_formatting.R: reads the data_for_macropru_analysis.csv file and reformats it for the LVR regressions in Section 7.1. This script produces the lvr_data_for_analysis.csv file to be read by Stata.
• plots_for_paper.R: reads the data_for_macropru_analysis.csv data file and produces tables of data (on screen) that are used for the figures in the paper.

The ‘Regression code’ folder contains Stata scripts, and csv data files that the Stata scripts read and write. The Stata scripts are:

• RDP_variable_definitions_code.do: reads the data_for_macropru_analysis.csv data file and formats it into regression variables.
• RDP_main_regressions.do: is run directly after running the RDP_variable_definitions_code.do script. It produces results for most of the regressions. It also produces the agg_analysis_rolling_regressions.csv data file containing data for most of the plots in Section 7.
• RDP_placebo_regressions.do: is run directly after running the RDP_variable_definitions_code.do script. It produces the placebo_results_Inv_policy.csv and placebo_results_IO_policy.csv data files for the Sections 5 and 6 placebo regression charts.
• LVR_regressions.do: reads the data from lvr_data_for_analysis.csv and produces the results in Section 7.1, on policy effects by LVR category.

Where possible, the data used to plot the figures will be released and can be found in ‘rdp-2021-07-graph-data.xlsx’. However, much of the data used is confidential or is restricted by 3rd party provider terms. The data restrictions on figures are:

• Figure 1 – Top two panels are not publically available (NPA) – confidentiality, bottom panel is publically available
• Figure 2 – NPA – confidentiality
• Figure 3 – NPA – confidentiality
• Figure 4 – NPA – confidentiality
• Figure 5 – NPA – confidentiality
• Figure 6 – NPA – confidentiality
• Figure 7 – NPA – confidentiality
• Figure 8 – NPA – confidentiality
• Figure 9 – NPA – confidentiality
• Figure 10 – Top two panels are publically available, bottom two panels are NPA – confidentiality
• Figure 11 – NPA – confidentiality
• Figure 12 – NPA – confidentiality
• Figure 13 – Publically available

Software versions

R version 4.0.3 and Stata version StataSE 16.1.

26 July 2021