Notes for replication of the results in:
"Transitional Dynamics of Dividend and Capital Gains Tax Cuts",
By Francois Gourio and Jianjun Miao.

(A) Baseline Model Results (Section 3)

Take all the m-files of the folder "Section 3" and install them in a sub-directory of your Matlab® folder. To replicate our main results, you will need to run the program divTransitionRun.m

The parameters are set in the file divTransitionParamFinal.m

The program divTransitionRun.m computes both the steady-states (option: compss=1) and the transitional dynamics (option: comptran=1). Note that the exact path for future taxes is set in the program on lines 135-154 and can easily be changed.

The program saves the intermediate results (files SSxxx.mat for the steady-states and restransitionxxx.mat for the transitional dynamics). Hence, if you are not changing the initial and final steady-states, but only the tax experiment, you do not need to re-compute the steady-states.

The overall computation takes some time, but all the results should be obtained in less than 10 hours on a recent desktop computer. You may need to have about 3Go of RAM since some large matrices are stored.

The file divTransitionPlotsAggNewNewCAPGAIN2.m produces the figures of the paper.
 
The file divTransitionPlotsAggNewNewCAPGAIN2withdata presents the figures together with the data.

We also include the .mat files with the results for the transitional dynamics in a folder "Matfiles". These files can, of course, be generated by the program divTransitionRun.m

Note: our programs use some programs written by John Cochrane (texprint) and
Joao Gomes (QuadNorm), which these researchers have made available on their web pages.

(B) Extended model with debt results (Section 4)

The programs are similar in organization. However, because of the interpolation, the results are not always precise and care must be taken to check that the computation of the steady-state and transitional dynamics converges well.

The main program is RunTransitionDebtGEMarch.m, which is organized in a way similar to divTransitionRun.m

The parameters are set in the file divTransitionDebtGEparamMARCH.m


(C) Data

The data construction is discussed in Appendix. The Stata® program used are included:

- First, run “cleandatanewJJupdate.do" to perform some basic data cleaning;
- Second, run “distributionfirmsMarchFinal3.do”. This program will extract a file with the series that we used to compare to the model.
- This file (named eqiss7209finaldataN5.out), is read by the Matlab file divGEdatadistributionnewMARCHvariation2.m
- The file eqiss7209finaldataN5.out is included.
- Note that to run the Stata file, you will need an extract of Compustat with the appropriate items (See our Stata programs for the variables used; we used the new Compustat presentation where the variables names are changed, e.g. instead of data12 we have sale). The Compustat data are  not included because of the size of the file and because Compustat data are proprietary.

See also our other paper in AEAJ Macro, and its programs which are used for the computation of cross-sectional moments. These programs are posted on: http://people.bu.edu/fgourio/research.html

(D) Contact information

Please report any problem or error in these programs to Francois Gourio or Jianjun Miao:
Email: fgourio@bu.edu OR miaoj@bu.edu
Address: Dept of Economics, Boston University, 270 Bay State Road, Boston MA 02215
Phone: 617-353-4534 and 617-353-6675