The folders “..\Sectoral_KL”, “..\Firm_KL”, and “..\EconWide_KL” contain the Matlab
codes to solve the models in Carvalho and Nechio (2016). In particular:
- “..\Sectoral_KL” contains the codes to solve the version of the model with sector-
specific factor markets presented in Section 2.1.
- “..\Firm_KL” contains the codes to solve the version of the model with firm-
specific factor markets presented in Section 2.2.
- “..\EconWide_KL” contains the codes to solve the version of the model with
economy-wide factor markets presented in Section 2.3.

Solving the models:

- Main_code.m solves the models that assume a path for nominal aggregate demand
   and produces the impulse response functions of Figure 1.
- Main_code_TR.m solves the models that assume monetary policy follows a
   Taylor rule and produces the impulse response functions of Figure 2.
- The codes Main_code.m and Main_code_TR.m solve the model using Chris
   Sims’ Gensys algorithm, whose files are contained in folder “..\Gensys”. For the
        benchmark cases with 67-sector economies, the two codes also simulate data for
       output and investment and tests for the calibration of the investment cost
      adjustment parameter (kappa).
- The models can be solved for 67-sector economies. As inputs, they require the file
   “falpha.csv”, which contains the distribution of price stickiness for the 67-sector
      economy (the order of the columns is: sectoral weight, sectoral frequency of price
     changes). This file is loaded by Main_code and Main_code_TR.m in order to
    solve the model. It should not be altered. The calculation that yields the
   distribution, based on the statistics from Nakamura and Steinsson (2008), is in the
  Excel file “DataFiveFactsPaper_AggregELI.xls”, available in the folder
     “..\DataFiles”.
- Models with economy-wide factor markets (in folder “..\EconWide _KL”) should
       take around 7 seconds to run. Models with sector-specific factor markets (in
      folder “..\Sectoral_KL”) should take around 96 seconds to run. Models with firm-
         specific factor markets (in folder “..\Firm_KL”) should take around 529 seconds
            to run.
- All codes were written using Matlab 12a.

Other files/functions in the folders:

- The files Het_Open_mats_1.mat, Het_Open_mats_3.mat and
Het_Open_mats_67.mat in each of the three folders (“..\Sectoral_KL”,
“..\Firm_KL”, and “..\EconWide_KL”) contain the matrices required for the
solution of the models that assume a path for nominal aggregate demand. The
matrices correspond to the models with one-sector, 3-sector and 67-sector
economies, respectively. They should not be altered.
- The file Het_Open_mats_TR_67.mat in each of the three folders
   (“..\Sectoral_KL”, “..\Firm_KL”, and “..\EconWide_KL”) contains the matrix
              required for the solution of the model that assume a Taylor rule. The matrix
             correspond to the models with 67-sector economies. They should not be altered.
- The folders “..\Sectoral_KL”, “..\Firm_KL”, and “..\EconWide_KL” also include
               the code “hptrend.m” which runs an Hodrick–Prescott filter on the model
                    simulated data to test for the calibration of the investment cost adjustment
                   parameter (kappa). The parameter kappa specified in the codes Main_code.m and
                  Main_code_TR.m is already calibrated to match the standard deviation of
                 investment in the data relative to the standard deviation of GDP when the number
                of sectors equal 67. The codes also run for other choices for kappa, but the
               relative standard deviation will not match the empirical evidence.
- The folder “..\Firm_KL” also contains the files “MainSolveCapital.m”,
           “ConditionsWoodford.m” and “ConditionsWoodford_noconstraint.m”, which are
                  used to numerically obtain the Phillips curve coefficients in the firm-specific
                 model. This minimization problem also requires minimization codes from Chris
                Sims contained in the folder “..\Optimize”, and the Matlab toolbox contained in
                   the folder “..\compecon”.

Replicating the paper figures:

- To replicate Figure 1, run Main_code.m from “..\EconWide_KL”,
       “..\Sectoral_KL”, and “..\Firm_KL”, in this order.
- To replicate Figure 2, run Main_code_TR.m from “..\EconWide_KL”,
       “..\Sectoral_KL”, and “..\Firm_KL”, in this order.
- To replicate Appendix Figure A1, in Main_code.m from “..\EconWide_KL”,
       “..\Sectoral_KL”, and “..\Firm_KL” set return to labor equal to one (chi=1), and
              run the codes for 67-sector and one-sector economies. The figure plots the
             variable called “kappa_tilde”. The model solution with one-sector economies uses
                Het_Open_mats_1.mat as input. All versions use falpha.csv as input.
- To replicate Appendix Figure A2, run Main_code.m from “..\EconWide_KL”,
       “..\Sectoral_KL”, and “..\Firm_KL” with return to labor equal to 2/3 (chi=2/3),
              with 67-sector economies. The figure plots the variable called “kappa_tilde”.
- To replicate Appendix Figure A3, run Main_code.m from “..\EconWide_KL”,
       “..\Sectoral_KL”, and “..\Firm_KL” for the 3-sector hypothetical economies with
              flexible and sticky price distributions. The flexible economies use as input
             falpha_flex.csv, while the sticky economies use as input falpha_sticky.csv. The
            panels in the figure plot the variable called “GDPt”.