Firm markups and firm patents

This project reproduces individual Compustat firms markups using the method of De Loecker, J., Eeckhout, J., & Unger, G. (2020). The rise of market power and the macroeconomic implications. The Quarterly Journal of Economics and matches firms to USPTO patents based on the crosswalk from Dorn, D., Hanson, G. H., Pisano, G., & Shu, P. (2020). Foreign competition and domestic innovation: Evidence from US patents. American Economic Review: Insights

Data

Some data is available on the repo, some has to be downloaded yourself because they are either too large or not public (compustat). Here is the list of files that must be in the data folder

theta_ALLsectors.dta [included] Sector elasticities estimations from Jan de Loecker website
theta_W_s_window.dta [included] Sector-year Elasticities estimations from Jan de Loecker website
macro_vars.dta [included] Macro variables for deflated measures from Jan de Loecker website
cw_patent_compustat_adhps.dta [NOT INCLUDED] Crosswalk table between patents and compustat ID from David Dorn website (link [I1])
patent_assignee.tsv [NOT INCLUDED] Patent to assignee unique ID from PatentsView
compustat_northamerica_annual.csv [NOT INCLUDED] Compustat - Capital IQ data
- Entire database, annual data, consolidated account.
- Note: The file name must match, don't forget to rename!

Reproduce

R is needed.

Run make in the root directory.

Results

Two tables are available in the out directory

markup

This table partially reproduces the intermediate table in the code of De Loecker & al. Definitions of variable are identical, and the reader may want to investigate their code for more information.

The estimations mu_0...mu_11 are based on various estimations of input elasticity.

markups.csv

   year         Compustat fiscal year
   gvkey        Compustat ID
   sale_D       Deflated sales

   /* Markup estimations based on costshare */
   mu_0         markup; costshare = 0.85 (fig 1 NBER)
   mu_1         markup; costshare = cogs_D/(cogs_D+kexp)
   mu_2         markup; costshare = cogs_D/(cogs_D+xsga_D+kexp) 

   /* Markup estimations based on median costshare by industry */ 
   mu_3         markup; 2 digit NAICS; costshare = cogs_D/(cogs_D+kexp)
   mu_4         markup; 3 digit NAICS; costshare = cogs_D/(cogs_D+kexp)
   mu_5         markup; 4 digits NAICS; costshare = cogs_D/(cogs_D+kexp)
   mu_6         markup; 2 digits NAICS; costshare = cogs_D/(cogs_D+kexp+xsga_D)
   mu_7         markup; 3 digits NAICS; costshare = cogs_D/(cogs_D+kexp+xsga_D)
   mu_8         markup; 4 digits NAICS; costshare = cogs_D/(cogs_D+kexp+xsga_D)
   
   /* Markup estimations based on production function estimate */
   mu_9         markup; estimated theta; sector theta
   mu_10        markup; estimated theta; sector-time theta (Benchmark)
   mu_11        markup; estimated theta; sector-time theta, alternative specification

cw_patent_compustat

This is a many-to-many table linking patents to compustat ID and assignee ID. One patent may have multiple assignee. gvkey_adhps is the original match from Dorn et al. . I try to match additional patents by leveraging patentsView disambiguation. The imputed (new) compustat ID is given by gvkey. A matching fit is also given by impscore in ]0,1].