Replication files for "The Impact of Online Competition on Local Newspapers: Evidence from the Introduction of Craigslist"

This repository contains replication code and data for Djourelova, Durante, and Martin (2023), "The Impact of Online Competition on Local Newspapers."

How to use this archive

1. R and Stata scripts in data_construction construct the analysis datasets from the raw data. Some parts of these scripts depend on proprietary data not included in the archive (see below for description). These scripts must be run in the order specified in the file name: all scripts beginning with "00_" must be run before scripts beginning with "01_", and so on. Scripts with the same number prefix may be run in any order. This step can be skipped, as the resulting processed datasets are included in the data folder.

2. R, Stata scripts in analysis run the analyses and produce tables and figures in the paper from the processed data. Processed datasets are included in the archive in the data folder. These scripts may be run in any order. See below for indices describing which file produces each table or figure in the manuscript and appendices. Some of the tables / figures also depend on proprietary data. These are noted in the indices.

Computational Requirements and Dependencies:

The code was last run on a 4-core Intel-based laptop with Windows 11 Pro. The code took less than 2 hours to execute (excluding the processing of proprietary datasets).

Stata dependencies (code last run with version 18 SE): reghdfe, distinct, did_mutiplegt, binscatter, zscore, regsave, estout, eclplot, shp2dta, geoinpoly, erepost (all as of Jan 2024)

Manuscript Figures / Tables to Source Code Index

Table / Figure	Code to Produce	Requires Proprietary Data?
Figure 1	make_maps.R
Figure 2	___do_Figures_Main.do
Figure 3	___do_Figures_Main.do
Figure 4	___do_Figures_Main.do	Yes
Figure 5	___do_Figures_Main.do
Table 1	___do_Tables_Main.do
Table 2	___do_Tables_Main.do
Table 3	___do_Tables_Main.do
Table 4	___do_Tables_Main.do
Table 5	___do_Tables_Main.do
Table 6	___do_Tables_Main.do	Yes
Table 7	___do_Tables_Main.do

Appendix Figures / Tables to Source Code Index

Table / Figure	Code to Produce	Requires Proprietary Data?
Figure A.1	___do_Tables_Appendix.do
Figure A.2	___do_Tables_Appendix.do
Figure B.1	N/A
Figure B.2	N/A
Figure B.3	N/A
Figure B.4	___do_Figures_Appendix.do	Yes
Figure B.5	___do_Figures_Appendix.do	Yes
Figure B.6	N/A
Figure B.7	N/A
Table A.1	___do_Tables.do
Table A.2	___do_Tables_Appendix.do	Yes (panel c)
Table A.3	___do_Tables_Appendix.do
Table A.4	___do_Tables_Appendix.do
Table A.5	___do_Tables_Appendix.do
Table A.6	___do_Tables_Appendix.do
Table A.7	___do_Tables_Appendix.do
Table A.8	___do_Tables_Appendix.do
Table A.9	___do_Tables_Appendix.do
Table A.10	___do_Tables_Appendix.do
Table A.11	___do_Tables_Appendix.do
Table A.12	___do_Tables_Appendix.do
Table A.13	___do_Tables_Appendix.do	Yes
Table A.14	___do_Tables_Appendix.do	Yes
Table A.15	___do_Tables_Appendix.do
Table A.16	___do_Tables_Appendix.do
Table A.17	___do_Tables_Appendix.do
Table A.18	___do_Tables_Appendix.do
Table A.19	___do_Tables_Appendix.do
Table A.20	___do_Tables_Appendix.do
Table B.1	___do_Figures_Main.do
Table B.2	___do_Tables_Appendix.do
Table B.3	data_construction/01_Topic Model.ipynb	Yes

Description of Data Files

The data directory contains the main analysis data sets, master_data_county_level.dta and master_data_newspaper_level.dta used to run the regressions. It also contains raw data in the following directories:

Directory	Description
_co99_d00_shp	County shape files
_counties_to_CDs	Crosswalks from county to congressional district
_counties_to_DMAs	Crosswalks from county to media market (DMA)
_zipcode	Zipcode to county crosswalks
annenberg	Predicted values of interest in news and classified sections given reader demographics, estimated in the National Annenberg Election Survey data (raw data not included)
census	Data from the US census used to crosswalk census tracts to counties, and for intercensal population estimates at county level
Classified_Prices	Digitized data from SRDS on classified rates at the newspaper-year level
Comscore	Aggregated website visits recorded by Comscore (raw data not included)
controls	Data for control variables used in regressions
craigslist_expansion	Data on opening dates of local Craigslist sites
`controls'	Data on county-level socio-economic characteristics from the US census and BLS
E&P	Digitized data from the Editor and Publisher yearbooks
GfK-MRI	Fitted models and predicted values of interest in news and classified sections given reader demographics, estimated in the GfK-MRI data (raw data not included)
ISPs	Data from the FCC on the number of Internet service providers by zip-code
Newspapers.com	Data on the number of classified pages per issue at the newspaper-year level, extracted from the Newspapers.com archive
Newspapers_content	Keyword counts and topic models extracted from the NewsBank database of newspaper content
political	Electoral data at county and county x congressional district level, along with information on members of congress used to generate keyword searches
population	County-level population data

List of Data Construction Scripts

Script	Requires Proprietary Data?
00_classifieds_clean_newspapers_com_data.R	Yes
00_clean_Annenberg.do	Yes
00_clean_Comscore.do	Yes
00_clean_ISPs.do
00_congress_outcomes.R
00_gfk_clean_gfkmri.R	Yes
00_keywordcounts_export_congressman_list.R	Yes
00_keywordcounts_export_nb_search_folders.R	Yes
00_turnout_splitvote_outcomes.R
01_gfk_predict_classified_interest.R	Yes
01_keywordcounts_export_keyword_mentions.R	Yes
01_Topic Model.ipynb	Yes
02__do_master_county_level.do
02__do_master_newspaper_level.do

Sources for Public Data Sets

1. Editor & Publisher yearbooks (E&P, 2010). We digitized the print yearbook editions for the years 1995-2010 using OCR software. Copies of the yearbooks can be located at https://www.worldcat.org/title/2210361. The dataset including the variables used in the analysis is provided in this package.

2. Standard Rate and Data Service: Newspaper Advertising Source (SRDS, 2006). We digitized the print editions of the yearbooks for the years 1995-2006 using a data transcription service. Copies of the yearbooks can be located at https://search.worldcat.org/title/56675270. The dataset including the varaiables used in the analysis is provided in this package.

3. Number of Internet Service Providers by zipcode (FCC, 2008). Publicly available at https://www.fcc.gov/form-477-data-zip-codes-number-high-speed-service-providers, and included in this package.

4. Database on Ideology, Money in Politics, and Election (BONICA, 2016). Publicly available at XXXXXXXXXXXXXXXXXXXXXXXXXXXXX , and included in this package.

5. Turnout in House, Senate and Presidential elections from XXXXXXXXXXXXXXXXX . XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX

6. County-level socio-economic characteristcis from the 2000 US Census (US Census Bureau, 2000). Publicly available at https://data.census.gov/, and included in this package.

7. County-level unemployment rates from the Bureau of Labor Statistics (BLS, 2000). Publicly available at https://www.bls.gov/lau/laucnty00.txt, and included in this package.

8. Population estimates by county and year from the National Center for Health Statistics (NCHS, 2010). Publicly available at https://www.cdc.gov/nchs/nvss/ bridged_race.htm.

Sources for Proprietary Data Sets

There are four proprietary or access-restricted data sets used in the paper. These are:

1. NewsBank full-text archive (NEWSBANK, 2010). These data are required to run the scripts in data_construction/newsbank_server, which generate counts of mentions of politician names and other keywords in newspaper text. We licensed the NewsBank data through the Stanford University Libraries. Contact jmcdowell@newsbank.com for licensing inquiries.

2. GfK-MRI Survey of the American Consumer (GFK-MRI, 2010). These data are used in the analysis of self-reported newspaper reading (reported in Table 6(a)). Contact Adriane.Heimann@mrisimmons.com for licensing inquiries.

3. National Annenberg Election Survey (NAES, 2008) restricted data. These data are used in the analysis of self-reported newspaper reading (reported in Table 6(b)). Visit the NAES home page to request access.

4. Comscore WRDS web traffic panel (COMSCORE, 2010). These data are used in the analysis of visits to the Craigslist.org domain (Figure 4). Visit https://wrds-www.wharton.upenn.edu/pages/about/data-vendors/comscore/ for licensing inquiries.

Data References

BONICA, A. (2016): “Database on Ideology, Money in Politics, and Elections,” Stanford, CA: Stanford University Libraries. https://data.stanford.edu/dime.

Bureau of Labor Statistics (2000): “Labor Force Data by County, 2000 Annual Averages,” https://www.bls.gov/lau/laucnty00.txt.

COMSCORE (2010): “WRDS ComScore,” Wharton Research Data Services. https://wrds-www.wharton. upenn.edu/pages/about/data-vendors/comscore/, accessed April, 2023.

U.S. Census Bureau (2000): “Decennial Census data release by county,” Retrieved from https://factfinder.census.gov.

E&P (2010): Editor & Publisher International Year Book, 1995-2010.

FCC (2008): “Federal Communications Commission Form 477 Additional Data, 1999-2008,” https://www. fcc.gov/general/fcc-form-477-additional-data, accessed April, 2023.

GFK-MRI (2010): “MRI-Simmons: Survey of the American Consumer 1999-2010,” GfK - Mediamark Research Intelligence. https://www.mrisimmons.com/our-data/national-studies/ survey-american-consumer/.

NAES (2008): “Annenberg Public Policy Center: National Annenberg Election Survey 2000, 2004, 2008,” University of Pennsylvania. https://www.annenbergpublicpolicycenter.org/ 2008-naes-telephone-and-online-data-sets/, accessed April, 2023.

NEWSBANK (2010): “Access World News Database,” https://www.newsbank.com/libraries/military/ solutions/access-world-news.

NCHS (2010): National Center for Health Statistics and U.S. Census Bureau “Vintage 2010 postcensal estimates of the resident population of the United States,” https://nchs/nvss/bridged_race.

SRDS (2006): Standard Rate and Data Service: Newspaper Advertising Source, 1995-2006.