NBER Working Papers and Chapters Metadata

This record contains commonly requested metadata for the NBER Working Paper series in various delimited UTF-8 or Stata .dta formats suitable for analysis by statistical or general purpose software. It can also load into spreadsheets with a single click. These files are intended for statistical analysis. Several formats for bibliographic software are linked from the NBER Information for Libraries page.

Available formats:

• csv
• dta
• tsv

All of these files should update automatically each week.

For the character files the first column is always the Working Paper number, then a tab, then the listed variable(s). The first row has variable names. Subordinate tables, like working_papers_authors or working_papers_jels, have multiple rows per paper, often organized by an order_num field. So the row in working_papers_authors with order_num = 0 is the first author, order_num = 1 is the second author, etc.

These files have one observation per paper, or one observation for each paper-author combination, paper-program or paper-jel combination. File names and content for the tab delimited versions:

There is also ref.tsv which combines paper number, author, title and issue date in a single line like a journal reference. In this file the author's names for each paper are concatenated into a single string. Variable names by file:
 

Clicking on the file name in the tsv subdirectory will open the file in a spreadsheet, for most users. Stata versions are in ./dta. Here is an example of a Stata program that prepares a bibliography of all papers in the Health Economics program:

Sample Stata code to list all papers in the HE program:

use "http://data.nber.org/nber_paper_chapter_metadata/dta/ref.dta"
keep paper author title
save ref,replace
clear
use "http://data.nber.org/nber_paper_chapter_metadata/dta/prog.dta"
keep if program=="EH"
merge 1:1 paper using ref 
list in 1/3

Working paper numbers have 4 or 5 digits and usually begin with a "w." Example "w12345". However, in years past the prefix letter could be an "h" (for Historical Paper series) or a "t" (for Technical Paper series). Both of these series are no longer distributed as new papers in favor of the single "w" coded Working Paper series. 

The published.tsv file shows where working papers were eventually published. It is not comprehensive. It combines information from authors, Crossref, RePEc and online vitae and journal names may vary by source.

Questions should be directed to:

Daniel Feenberg
feenberg@nber.org
+1 617-682-6204

Alternative format

A different extract with slightly different fields and parameters is available under

• tab

and as the gzipped files under

• csv

The files in those directories include dumps from our canonical database for tracking working papers and chapters, with some private information redacted.

The gzipped files (*.csv.gz) use alternative characters as the comma and quote characters, hex characters BF and AC.

The working_papers table is the main table, with one row per paper. Chapters in NBER books are also listed in that table, with ID numbers with 'c' in front.

Other subordinate tables, like working_papers_authors or working_papers_jels, have multiple rows per paper, often organized by an order_num field. So the row in working_papers_authors with order_num = 0 is the first author, order_num = 1 is the second author, etc.