DynareR: A Seamless Integration of R and Dynare

Sagiru Mati

2022-04-30

About DynareR

DynareR is an R package that can run Dynare program from R Markdown.

Requirements

Users need the following in order to knit this document:

  1. Dynare 4.6.1 or above

  2. Octave 5.2.0 or above

  3. Dynare is installed in the standard location as follows:

If dynare and Octave are installed in standard location, DynareR package will take care of the configurations, which include adding matlab directory to path, using the latest installed dynare and so on. Otherwise, users have to specify the matlab folder using add_path function, set the Octave path using the set_octave_path function, or set dynare version using the set_dynare_version function.

Installation

DynareR can be installed using the following commands in R.

install.packages("DynareR")

          OR
          
devtools::install_github('sagirumati/DynareR')

Usage

Please load the DynareR package as follows:

```{r DynareR}                                                             
library(DynareR)
```

Then create a chunk for dynare (adopted from Dynare example file bkk) as shown below:

```{dynare bkk,eval=T} 
/*
 * This file implements the multi-country RBC model with time to build,
 * described in Backus, Kehoe and Kydland (1992): "International Real Business
 * Cycles", Journal of Political Economy, 100(4), 745-775.
 *
 * The notation for the variable names are the same in this file than in the paper.
 * However the timing convention is different: we had to taken into account the
 * fact that in Dynare, if a variable is denoted at the current period, then
 * this variable must be also decided at the current period.
 * Concretely, here are the differences between the paper and the model file:
 * - z_t in the model file is equal to z_{t+1} in the paper
 * - k_t in the model file is equal to k_{t+J} in the paper
 * - s_t in the model file is equal to s_{J,t}=s_{J-1,t+1}=...=s_{1,t+J-1} in the paper
 *
 * The macroprocessor is used in this file to create a loop over countries.
 * Only two countries are used here (as in the paper), but it is easy to add
 * new countries in the corresponding macro-variable and completing the
 * calibration.
 *
 * The calibration is the same than in the paper. The results in terms of
 * moments of variables are very close to that of the paper (but not equal
 * since the authors a different solution method).
 *
 * This implementation was written by Sebastien Villemot. Please note that the
 * following copyright notice only applies to this Dynare implementation of the
 * model.
 */

/*
 * Copyright (C) 2010 Dynare Team
 *
 * This file is part of Dynare.
 *
 * Dynare is free software: you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation, either version 3 of the License, or
 * (at your option) any later version.
 *
 * Dynare is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with Dynare.  If not, see <http://www.gnu.org/licenses/>.
 */

@#define countries = [ "H", "F" ]
@#define J = 4

@#for co in countries
var C_@{co} L_@{co} N_@{co} A_@{co} K_@{co} Z_@{co} X_@{co} LAMBDA_@{co} S_@{co} NX_@{co} Y_@{co};

varexo E_@{co};

parameters beta_@{co} alpha_@{co} eta_@{co} mu_@{co} gamma_@{co} theta_@{co} nu_@{co} sigma_@{co} delta_@{co} phi_@{co} psi_@{co} rho_@{co}_@{co};
@#endfor

// Lagrange multiplier of aggregate constraint
var LGM;

parameters rho_@{countries[1]}_@{countries[2]} rho_@{countries[2]}_@{countries[1]};

model;
@#for co in countries

Y_@{co} = ((LAMBDA_@{co}*K_@{co}(-@{J})^theta_@{co}*N_@{co}^(1-theta_@{co}))^(-nu_@{co}) + sigma_@{co}*Z_@{co}(-1)^(-nu_@{co}))^(-1/nu_@{co});
K_@{co} = (1-delta_@{co})*K_@{co}(-1) + S_@{co};
X_@{co} =
@# for lag in (-J+1):0
          + phi_@{co}*S_@{co}(@{lag})
@# endfor
;

A_@{co} = (1-eta_@{co})*A_@{co}(-1) + N_@{co};
L_@{co} = 1 - alpha_@{co}*N_@{co} - (1-alpha_@{co})*eta_@{co}*A_@{co}(-1);

// Utility multiplied by gamma
# U_@{co} = (C_@{co}^mu_@{co}*L_@{co}^(1-mu_@{co}))^gamma_@{co};

// FOC with respect to consumption
psi_@{co}*mu_@{co}/C_@{co}*U_@{co} = LGM;

// FOC with respect to labor
// NOTE: this condition is only valid for alpha = 1
psi_@{co}*(1-mu_@{co})/L_@{co}*U_@{co}*(-alpha_@{co}) = - LGM * (1-theta_@{co})/N_@{co}*(LAMBDA_@{co}*K_@{co}(-@{J})^theta_@{co}*N_@{co}^(1-theta_@{co}))^(-nu_@{co})*Y_@{co}^(1+nu_@{co});

// FOC with respect to capital
@# for lag in 0:(J-1)
 +beta_@{co}^@{lag}*LGM(+@{lag})*phi_@{co}
@# endfor
@# for lag in 1:J
 -beta_@{co}^@{lag}*LGM(+@{lag})*phi_@{co}*(1-delta_@{co})
@# endfor
 = beta_@{co}^@{J}*LGM(+@{J})*theta_@{co}/K_@{co}*(LAMBDA_@{co}(+@{J})*K_@{co}^theta_@{co}*N_@{co}(+@{J})^(1-theta_@{co}))^(-nu_@{co})*Y_@{co}(+@{J})^(1+nu_@{co});

// FOC with respect to stock of inventories
 LGM=beta_@{co}*LGM(+1)*(1+sigma_@{co}*Z_@{co}^(-nu_@{co}-1)*Y_@{co}(+1)^(1+nu_@{co}));

// Shock process
@# if co == countries[1]
@#  define alt_co = countries[2]
@# else
@#  define alt_co = countries[1]
@# endif
 (LAMBDA_@{co}-1) = rho_@{co}_@{co}*(LAMBDA_@{co}(-1)-1) + rho_@{co}_@{alt_co}*(LAMBDA_@{alt_co}(-1)-1) + E_@{co};


NX_@{co} = (Y_@{co} - (C_@{co} + X_@{co} + Z_@{co} - Z_@{co}(-1)))/Y_@{co};

@#endfor

// World ressource constraint
@#for co in countries
  +C_@{co} + X_@{co} + Z_@{co} - Z_@{co}(-1)
@#endfor
    =
@#for co in countries
  +Y_@{co}
@#endfor
    ;

end;

@#for co in countries
beta_@{co} = 0.99;
mu_@{co} = 0.34;
gamma_@{co} = -1.0;
alpha_@{co} = 1;
eta_@{co} = 0.5; // Irrelevant when alpha=1
theta_@{co} = 0.36;
nu_@{co} = 3;
sigma_@{co} = 0.01;
delta_@{co} = 0.025;
phi_@{co} = 1/@{J};
psi_@{co} = 0.5;
@#endfor

rho_H_H = 0.906;
rho_F_F = 0.906;
rho_H_F = 0.088;
rho_F_H = 0.088;

initval;
@#for co in countries
LAMBDA_@{co} = 1;
NX_@{co} = 0;
Z_@{co} = 1;
A_@{co} = 1;
L_@{co} = 0.5;
N_@{co} = 0.5;
Y_@{co} = 1;
K_@{co} = 1;
C_@{co} = 1;
S_@{co} = 1;
X_@{co} = 1;

E_@{co} = 0;
@#endfor

LGM = 1;
end;

shocks;
var E_H; stderr 0.00852;
var E_F; stderr 0.00852;
corr E_H, E_F = 0.258;
end;

steady;
check;

stoch_simul(order=1, hp_filter=1600);
```  

The above chunk creates a Dynare program with the chunk’s content, then automatically run Dynare, which will save Dynare outputs in the current directory.

Please note that DynareR uses the chunk name as the model name. So, the outpus of Dynare are saved in a folder with its respective chunk name. Thus a new folder bkk/ will be created in your current working directory.

By default, dynare chunk imports log output as a list of dataframes, which can be accessed via dynare$modelName. Therefore to access the outputs of the bkk model produced by the dynare chunk, use dynare$bkk.

Use inline code `r dynare$bkk$moments[2,3]` to access the value of second row and third column of the moments, which is .

Plotting the IRF

The Impulse Response Function (IRF) is saved by default in bkk/bkk/graphs/ folder with the IRF’s name bkk_IRF_E_H2.pdf, where bkk is the Dynare model’s name. Therefore, you need to add stoch_simul(graph_format = (pdf)) to change the default saving behaviour of Dynare from eps to pdf.

DynareR functions for base R

The DynareR package is also designed to work with base R. The following functions show how to work with DynareR outside the R Markdown or Quarto documents.

The include_IRF function

Use this function to embed the graphs Impulse Response Function (IRF) in R Markdown or Quarto document.

The Impulse Response Function (IRF) of the bkk model can be fetched using the following R chunk. Note that only the last part of the IRF’s name (E_H2) is needed, that is bkk_IRF_ is excluded. Also note that out.extra='trim={0cm 7cm 0cm 7cm},clip' is used to trim the white space above and below the IRF.

```{r IRF,out.extra='trim={0cm 7cm 0cm 7cm},clip',fig.cap="Another of figure generated from Dynare software"} 
include_IRF("bkk","E_H2")

# Alternatively, use the path argument 

```

include_IRF(model="bkk",IRF = "E_H2")

# Alternatively, use the path argument 

include_IRF(path="bkk/bkk/graphs/bkk_IRF_E_H2.pdf")

However, Dynare figure can only be dynamically included if the output format is pdf as Dynare produces pdf and eps graphs only.

The write_dyn function

This function writes a new dyn file.

Use write_dyn(code="code",model="someModel") if you want the Dynare file to live in the current working directory. Use write_dyn(code="code",model="path/to/someDirectory/someModel") if you want the Dynare file to live in the path different from the current working directory.

dynareCodes='var y, c, k, a, h, b;
varexo e, u;
parameters beta, rho, alpha, delta, theta, psi, tau;
alpha = 0.36;
rho   = 0.95;
tau   = 0.025;
beta  = 0.99;
delta = 0.025;
psi   = 0;
theta = 2.95;
phi   = 0.1;
model;
c*theta*h^(1+psi)=(1-alpha)*y;
k = beta*(((exp(b)*c)/(exp(b(+1))*c(+1)))
          *(exp(b(+1))*alpha*y(+1)+(1-delta)*k));
y = exp(a)*(k(-1)^alpha)*(h^(1-alpha));
k = exp(b)*(y-c)+(1-delta)*k(-1);
a = rho*a(-1)+tau*b(-1) + e;
b = tau*a(-1)+rho*b(-1) + u;
end;
initval;
y = 1.08068253095672;
c = 0.80359242014163;
h = 0.29175631001732;
k = 11.08360443260358;
a = 0;
b = 0;
e = 0;
u = 0;
end;

shocks;
var e; stderr 0.009;
var u; stderr 0.009;
var e, u = phi*0.009*0.009;
end;

stoch_simul;'


write_dyn(code=dynareCodes, model="example1")

write_dyn(code=dynareCodes,model="DynareR/write_dyn/example1")

The write_mod function

This function writes a new mod file.

Use write_mod(code="code",model="someModel") if you want the Dynare file to live in the current working directory. Use write_mod(code="code",model="path/to/someDirectory/someModel") if you want the Dynare file to live in the path different from the current working directory.

DynareCodes='var y, c, k, a, h, b;
varexo e, u;
parameters beta, rho, alpha, delta, theta, psi, tau;
alpha = 0.36;
rho   = 0.95;
tau   = 0.025;
beta  = 0.99;
delta = 0.025;
psi   = 0;
theta = 2.95;
phi   = 0.1;
model;
c*theta*h^(1+psi)=(1-alpha)*y;
k = beta*(((exp(b)*c)/(exp(b(+1))*c(+1)))
          *(exp(b(+1))*alpha*y(+1)+(1-delta)*k));
y = exp(a)*(k(-1)^alpha)*(h^(1-alpha));
k = exp(b)*(y-c)+(1-delta)*k(-1);
a = rho*a(-1)+tau*b(-1) + e;
b = tau*a(-1)+rho*b(-1) + u;
end;
initval;
y = 1.08068253095672;
c = 0.80359242014163;
h = 0.29175631001732;
k = 11.08360443260358;
a = 0;
b = 0;
e = 0;
u = 0;
end;

shocks;
var e; stderr 0.009;
var u; stderr 0.009;
var e, u = phi*0.009*0.009;
end;

stoch_simul;'


write_mod(model="example1",code=dynareCodes)

write_mod(code=dynareCodes,model="DynareR/write_mod/example1")

The run_dynare function

Create and run Dynare mod file

Use this function to create and run Dynare mod file. Use run_dynare(code="code",model="someModel") if you want the Dynare files to live in the current working directory. Use run_dynare(code="code",model="path/to/someDirectory/someModel") if you want the Dynare files to live in the path different from the current working directory. Use import_log=T argument to return the dynare log file as list of dataframes in an environment dynare, which can be accessed via dynare$modelName.

DynareCodes='var y, c, k, a, h, b;
varexo e, u;
parameters beta, rho, alpha, delta, theta, psi, tau;
alpha = 0.36;
rho   = 0.95;
tau   = 0.025;
beta  = 0.99;
delta = 0.025;
psi   = 0;
theta = 2.95;
phi   = 0.1;
model;
c*theta*h^(1+psi)=(1-alpha)*y;
k = beta*(((exp(b)*c)/(exp(b(+1))*c(+1)))
          *(exp(b(+1))*alpha*y(+1)+(1-delta)*k));
y = exp(a)*(k(-1)^alpha)*(h^(1-alpha));
k = exp(b)*(y-c)+(1-delta)*k(-1);
a = rho*a(-1)+tau*b(-1) + e;
b = tau*a(-1)+rho*b(-1) + u;
end;
initval;
y = 1.08068253095672;
c = 0.80359242014163;
h = 0.29175631001732;
k = 11.08360443260358;
a = 0;
b = 0;
e = 0;
u = 0;
end;

shocks;
var e; stderr 0.009;
var u; stderr 0.009;
var e, u = phi*0.009*0.009;
end;

stoch_simul;'

run_dynare(code=DynareCodes,model="example1",import_log = T)
run_dynare(code=DynareCodes,model="DynareR/run_dynare/example1")

The run_models function

Run multiple existing mod or dyn files.

Use this function to execute multiple existing Dynare files. Use run_models(model="someModel") if the Dynare files live in the current working directory. Use run_models(model="path/to/someDirectory/someModel") if the Dynare files live in the path different from the current working directory. Use run_models() to exectute all the dynare models in the current working directory. Use run_models("path/to/someDirectory*) to run all the dynare models in path/to/someDirectory.

Where agtrend.mod, bkk.mod and example1.mod are the Dynare model files (with mod or dyn extension), which live in the current working directory.


demo(agtrend)
demo(bkk)
demo(example1)

# Provide the list of the `Dynare` files in a vector
# Ensure that "agtrend.mod", "bkk.mod" and "example1.mod"
# live in the current working directory

# Copy the dynare files to the current working directory

lapply(c("agtrend","bkk","example1"),\(x) file.copy(paste0(x,"/",x,".mod"),"."))

run_models(c("agtrend","bkk","example1")) # Run the models in the vector.

To run all Dynare models that live in the current working directory, use the following:

run_models() # Run all models in Current Working Directory.

To run all Dynare models that live in particular path (for example ‘DynareR/run_dynare/’ folder), use the following:


# Copy the dynare files to the 'DynareR/run_dynare' directory

lapply(c("agtrend","bkk","example1"),\(x) file.copy(paste0(x,".mod"),"DynareR/run_dynare"))

run_models(model = 'DynareR/run_dynare*') # notice the * at the end

import_log function

This function returns the dynare log output as a list of dataframes, which include summary, shocks, policy, moments, decomposition, correlation and autocorrelation. The list is accessible via dynare$modelName. if the model name is bkk, the policy variables can be obtained via dynare$bkk$policy as a dataframe.

import_log(model="bkk")

import_log(path="bkk/bkk.log")

knitr::kable(dynare$bkk$autocorrelation)  %>% kableExtra::kable_styling(latex_options = c("basic","hold_position","scale_down")) %>%
 kableExtra::footnote(general="Some footnote with equation $\\alpha x^2+\\beta x+c=0$", general_title = "*",footnote_as_chunk=T,threeparttable=T,escape=F) %>%
kableExtra::row_spec(0,bold=T)

set_dynare_version function

On Windows, you can set the version of dynare you want to use. By default, DynareR package does this for you if the dynare version ranges from 4.6.1 to 9.9. However, if you are using the development version of dynare, for example version 6-unstable-2022-04-03-0800-700a0e3a, you can override the default as follows

set_dynare_version("6-unstable-2022-04-03-0800-700a0e3a")

set_octave_path function

You can use this function if Octave is not installed in the standard location

set_octave_path('C:/Program Files/GNU Octave/Octave-6.4.0/mingw64/bin/octave20.exe')

add_path function

This function is a wrapper of addpath in Octave. If dynare is not installed in the standard location, use this function to add the matlab subdirectory. By default, DynareR does this for if dynare is installed in the standard location.


add_path('/usr/lib/dynare/matlab')#  Default for Linux

add_path('c:/dynare/5.1/matlab') # Default for Windows, but 5.1 can change if later version of
# `Dynare` is installed.

add_path('/usr/lib/dynare/matlab') # Default for macOS

Demo

The demo files are included and can be accessed via demo(package=“DynareR”)

demo(run_dynare)
demo(run_models)
demo(import_log)

Template

Template for R Markdown is created. Go to file->New File->R Markdown-> From Template->DynareR.





Please download the example files from Github.