Orbis at NBER

We have split the files by country to reduce them to a reasonable size, and converted them to Stata .dta format. If you need other formats or have questions about the conversion process, including concerns that information may have been lost, please contact Daniel Feenberg.

Because these files are still quite large, we remind you that the Stata -use- statement can filter by variable and observation. So it isn't necessary to store any variables or observations you won't be using in your analysis.

You may notice that many of the Orbis variables are long strings with few possible values, such as the name of a stock exchange (given 71 bytes) The Stata -encode- command can reduce those to 1 or 2 bytes quite easily.

Documentation from BvD

• Ownership files (pdf)
• Variable list (xlsx)
• Global Format Definitions
• Online ORBIS system documentation

Local and other documentation

• FAQ
• Index of files by variable name
• Index of variables by file
• .dta file of variable names and characteristics
• .html file of record counts by file and country
• .dta file of record counts by file and country
• ISO Country
• .do file to create macro list of countries Codes
• Suggestions for dealing with large Stata files

Notes:

• Data cover 2007 through June 2019.
• Countries are referred to by 2 letter ISO country code.
• Paths are relative to /home/data/bvd/ on the NBER Linux cluster.
• Files with all countries are in ./world
• Files divided by country are in ./bycnty/CC/
• Stata .do files used in the conversion to .dta are in ./code
• Stata .dta files with variable information are in ./descsave
• Stata .log files from the conversion are in ./logs
• Variable and file names are all lowercase. Country directory names are uppercase.
• Most variable names are created by Stata from variable descriptions. In a few cases BvD provided names.
• BvD did not provide any additional documentation.
• A single copy of all the files in .dta format totals 2.3TB, but compresses to 254GB in our ZFS file store. Nevertheless, Stata decompresses the data in core. The largest single countries are Russia and the US, with 28GB and 20GB of compressed and ten times that amount in core.

Descriptive Data

• Most of these files had many stray double quote marks. Stata was instructed to ignore them.
• The original files are in ./descriptive/rar and ./descriptive/txt

Financials - Detailed cash flow and interim June text

• The original files are in ./detail/rar and ./detail/txt

Financials - Gobal format incl histo for industries June text

• Where the file description was the same as a file in the Detailed group, an "h" was prefixed to the name.
• The original files are in ./global/rar and ./global/txt

Ownership Histo June text

• Each ownership link is kept twice - once in the Ownership (Shareholder) Links file and once in the Subsidiary Links file according to the country code of the owner or subsidiary.
• The original files are in ./ownership/rar and ./ownership/txt