Supplementary Materials for ÒTaxing Atlas: Executive Compensation, Firm Size, and Their Impact on Optimal Top Income Tax RatesÓ 
by Laurence Ales, Antonio AndrŽs Bellofatto and Jessie Jiaxu Wang
January 2017


This archive contains the files for replicating the numerical results in the paper.
Data computations are performed in Stata. The rest of the codes are written in Matlab. In addition, computations connected to optimal firm size distortions (in Appendix G) require GPOPS-ii optimal control software for Matlab, as well as the SNOPT optimization software.

The specific environment which generates the computation times reported below is the following:

Mac Pro (Late 2013)
3.7 GHz Quad-Core Intel Xeon E5
16 GB 1866 MHz DDR3 ECC
OS X El Capitan 10.11.1
Stata/SE 13.1 for Mac (64-bit Intel)
Matlab R2015a (64-bit) 
GPOPS-ii version 2.2 using SNOPT

To replicate the numerical results in the paper, follow these steps (note that directory paths should be modified appropriately):


0. DATA PREPARATION (time to run: 1 minute approx.)
	
	0.i  In /Data/US run do-compustat-US.do. 
	This processes Compustat data in Sale_emp.dta and CPI data in cpi.xlsx, and produces US.dta.

	0.ii In /Data/CrossCountry run do-compustat-canada.do and do-compustat-global.do. 
	The former processes Compustat North America data in Sale_emp_canada.dta and CPI data in cpi.xlsx, and generates Canada.dta. 
	The latter processes Compustat Global data in Sale_emp_global.dta and CPI data in cpi.xlsx, and generates Global.dta.
 

1. ESTIMATE THE ELASTICITY OF FIRM SIZE WITH RESPECT TO SALES (time to run: less than 1 min.)

	1.i   In /Results run do-US-table1.do. 
	This uses US.dta as input and generates the values in Table 1 in the paper.
	The benchmark elasticity of firm size with respect to sales (US case) is saved as estim_elasticity.csv.

	1.ii  In /Results run do-US-robustManagers.do. 
	This uses US.dta as input and generates the data to create Figure 1 in the paper. Such data is stored as estim_elasticity_bymanagement.csv. 

	1.iii In /Results run do-canada.do. 
	This uses Canada.dta to estimate the elasticity of firm size with respect to sales for Canada. Such elasticity is saved as estim_elasticity_canada.csv.

	1.iv  In /Results run do-global.do. 
	This uses Global.dta to estimate the elasticity of firm size with respect to sales for non-US countries, except for Canada.  The result is saved as estim_elasticity_bycountry.csv.


2. COMPUTE OPTIMAL TOP TAXES (time to run: less than 1 min.)

In /Results run Master_TopTaxes.m. 
This generates Figures 2, 3 and 4, Table 2, Table 3, and the values in Table 4 (the OECD data showing up in the last column of Table 4 is contained in the Excel sheet OECD_Statutory_Comparison.xlsx).
The inputs for this step are: the elasticities estimated in step 1 above, cross-country data on the Pareto tail index for sales contained in /Data/CrossCountry/ay_data.csv (taken from di Giovanni and Levchenko (2013)), and cross-country data on the elasticity of managerial compensation wrt sales in /Data/CrossCountry/robertsLaw.dta (kindly shared by Miguel Ferreira).


3. COMPUTE RESULTS IN APPENDICES (time to run: 2-3 min.)

	3.i In /Results/Appendices run Appendix_F.m. 
	This performs sensitivity analyses on the elasticity of firm size wrt sales, and on the Pareto tail parameter of the firm size distribution. 
	Those results are reported in Table 5 in Appendix F.

	3.ii Replicating Appendix G first requires downloading the files plogncdf.m and plognpdf.m created by Ben Lockwood (Wharton) available at: https://github.com/bblockwood/utilities/tree/master/plogn. These files should be placed in the directory /Results/Appendices/Appendix_G. Once this is done, in /Results/Appendices/Appendix_G run the following Matlab files:
		* Appendix_G_code1.m
		* Appendix_G_code2.m
		* Appendix_G_code3.m
	These files generate, respectively, the values for firm size taxes in rows 1-3 of Table 6 (in Appendix G).