From 31f501ab811e6a77e98bc252c9d9be35d8d60bdd Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Tomasz=20Wo=C5=BAniak?= Date: Fri, 2 Aug 2024 01:50:49 +1000 Subject: [PATCH] the last --- index-speaker.html | 120 +++++++++++++++++++++++++---------------- index.html | 120 +++++++++++++++++++++++++---------------- index.qmd | 23 ++++++++ w4UKR_bsvars_Granger.R | 19 +++++++ 4 files changed, 188 insertions(+), 94 deletions(-) diff --git a/index-speaker.html b/index-speaker.html index 6f1e40a..0d737d4 100644 --- a/index-speaker.html +++ b/index-speaker.html @@ -536,6 +536,23 @@

b

- progress bar

+
+

My first steps with bsvars

+

Exercise.

+
    +
  1. Download R script w4UKR_bsvars.R
    2. +
  2. Open the file in your RStudio
    3. +
  3. Execute the code line by line
    4. +
+

You have just estimated your first model using the bsvars package! YAY!

+
    +
  1. Modify the line specifying the model to
    2. +
+
specify_bsvar_t$new(p = 4) |>
+
    +
  1. Run the code line by line
    2. +
+

bsvars models and identification

@@ -829,6 +846,15 @@

…for autoregressive parameters.

  • use the verify_autoregression() function
    • +
    +

    bsvars hypothesis verification

    +

    …for autoregressive parameters.

    +

    Exercise.

    +
      +
    1. Download the R script w4UKR_bsvars_Granger.R
      2. +
    2. Verify the hypothesis of no Granger causality from ttr to gdp
      3. +
    +

    bsvars modeling of monetary policy

    @@ -903,36 +929,36 @@

    Identified shocks.

    bsvars modeling of monetary policy

    Specify and estimate a lower-triangular SVAR-SV.

    -
    library(bsvars)
-load("soe.rda")
-soe           = as.matrix(soe)
-
-TT            = nrow(soe)
-lag_order     = 8
-lag_exogenous = 4
-T             = TT - max(lag_order, lag_exogenous)
-
-exogenous     = matrix(NA, TT - lag_exogenous, 0)
-for (i in 0:lag_exogenous) {
-  exogenous   = cbind(exogenous, as.matrix(soe[(lag_exogenous - i + 1):(TT - i), 6:8]))
-}
-
-set.seed(1234)
-spec          = specify_bsvar_sv$new(
-  data        = tail(soe[,1:5], T),
-  p           = lag_order,
-  exogenous   = tail(exogenous, T)
-)
-
-burn          = estimate(spec, 1e4)
-post          = estimate(burn, 1e4)
    +
    library(bsvars)
+load("soe.rda")
+soe           = as.matrix(soe)
+
+TT            = nrow(soe)
+lag_order     = 8
+lag_exogenous = 4
+T             = TT - max(lag_order, lag_exogenous)
+
+exogenous     = matrix(NA, TT - lag_exogenous, 0)
+for (i in 0:lag_exogenous) {
+  exogenous   = cbind(exogenous, as.matrix(soe[(lag_exogenous - i + 1):(TT - i), 6:8]))
+}
+
+set.seed(1234)
+spec          = specify_bsvar_sv$new(
+  data        = tail(soe[,1:5], T),
+  p           = lag_order,
+  exogenous   = tail(exogenous, T)
+)
+
+burn          = estimate(spec, 1e4)
+post          = estimate(burn, 1e4)

    bsvars modeling of monetary policy

    SVAR-SV: Compute and plot impulse responses.

    -
    post |> compute_impulse_responses(horizon = 60) |> plot(probability = 0.68, col = "#0056B9")
    +
    post |> compute_impulse_responses(horizon = 60) |> plot(probability = 0.68, col = "#0056B9")
    @@ -940,7 +966,7 @@

    SVAR-SV: Compute and plot i

    bsvars modeling of monetary policy

    SVAR-SV: Forecast error variance decompositions.

    -
    post |> compute_variance_decompositions(horizon = 60) |> plot(col = bsvars_grad)
    +
    post |> compute_variance_decompositions(horizon = 60) |> plot(col = bsvars_grad)
    @@ -948,7 +974,7 @@

    SVAR-SV: Forecast error

    bsvars modeling of monetary policy

    SVAR-SV: Structural shocks.

    -
    post |> compute_structural_shocks() |> plot(col = "#0056B9")
    +
    post |> compute_structural_shocks() |> plot(col = "#0056B9")
    @@ -956,7 +982,7 @@

    SVAR-SV: Structural shocks.

    bsvars modeling of monetary policy

    SVAR-SV: Shocks’ conditional standard deviations.

    -
    post |> compute_conditional_sd() |> plot(col = "#0056B9")
    +
    post |> compute_conditional_sd() |> plot(col = "#0056B9")
    @@ -964,7 +990,7 @@

    SVAR-SV: Shocks’ cond

    bsvars modeling of monetary policy

    SVAR-SV: Homoskedasticity verification.

    -
    post |> verify_identification() |> summary()
    +
    post |> verify_identification() |> summary()
     **************************************************|
  bsvars: Bayesian Structural Vector Autoregressions|
@@ -988,7 +1014,7 @@ 
    SVAR-SV: Homoskedasticity verifi
 
    bsvars modeling of monetary policy
    
 
    SVAR-SV: Fitted Values.
    
 
    
-
    post |> compute_fitted_values() |> plot(col = "#0056B9")
    
+
    post |> compute_fitted_values() |> plot(col = "#0056B9")
    
 
 
    
 
@@ -996,9 +1022,9 @@ 
    SVAR-SV: Fitted Values.
    
 
    bsvars modeling of monetary policy
    
 
    SVAR-SV: Does foreign sector matter?
    
 
    
-
    A0 = matrix(NA, 5, 56)
-A0[,45:56] = 0
-post |> verify_autoregression(hypothesis = A0) |> summary()
    
+
    A0 = matrix(NA, 5, 56)
+A0[,45:56] = 0
+post |> verify_autoregression(hypothesis = A0) |> summary()
    
 
    
 
     **************************************************|
  bsvars: Bayesian Structural Vector Autoregressions|
@@ -1023,22 +1049,22 @@ 
    SVAR-SV: Does foreign sector matter?
 
    bsvars modeling of monetary policy
    
 
    Specify and estimate a lower-triangular SVAR-t.
    
 
    
-
    set.seed(1234)
-spec          = specify_bsvar_t$new(
-  data        = tail(soe[,1:5], T),
-  p           = lag_order,
-  exogenous   = tail(exogenous, T)
-)
-
-burn          = estimate(spec, 1e4)
-post          = estimate(burn, 1e4)
    
+
    set.seed(1234)
+spec          = specify_bsvar_t$new(
+  data        = tail(soe[,1:5], T),
+  p           = lag_order,
+  exogenous   = tail(exogenous, T)
+)
+
+burn          = estimate(spec, 1e4)
+post          = estimate(burn, 1e4)
    
 
    
 
 
    
 
    bsvars modeling of monetary policy
    
 
    SVAR-t: Compute and plot impulse responses.
    
 
    
-
    post |> compute_impulse_responses(horizon = 60) |> plot(probability = 0.68, col = "#0056B9" )
    
+
    post |> compute_impulse_responses(horizon = 60) |> plot(probability = 0.68, col = "#0056B9" )
    
 
 
    
 
    
@@ -1046,7 +1072,7 @@ 
    SVAR-t: Compute and plot imp
 
    bsvars modeling of monetary policy
    
 
    SVAR-t: Forecast error variance decompositions.
    
 
    
-
    post |> compute_variance_decompositions(horizon = 60) |> plot(col = bsvars_grad )
    
+
    post |> compute_variance_decompositions(horizon = 60) |> plot(col = bsvars_grad )
    
 
 
    
 
@@ -1054,7 +1080,7 @@ 
    SVAR-t: Forecast error v
 
    bsvars modeling of monetary policy
    
 
    SVAR-t: Structural shocks.
    
 
    
-
    post |> compute_structural_shocks() |> plot(col = "#0056B9" )
    
+
    post |> compute_structural_shocks() |> plot(col = "#0056B9" )
    
 
 
    
 
@@ -1062,7 +1088,7 @@ 
    SVAR-t: Structural shocks.
    
 
    bsvars modeling of monetary policy
    
 
    SVAR-t: Normality verification.
    
 
    
-
    post |> verify_identification() |> summary( )
    
+
    post |> verify_identification() |> summary( )
    
 
    
 
     **************************************************|
  bsvars: Bayesian Structural Vector Autoregressions|
@@ -1086,7 +1112,7 @@ 
    SVAR-t: Normality verification.
    
 
    bsvars modeling of monetary policy
    
 
    SVAR-t: Degrees-of-freedom posterior density.
    
 
    
-
    post$posterior$df |> hist(breaks = 100, col = "#0056B9", border = "#0056B9", bty = "n" )
    
+
    post$posterior$df |> hist(breaks = 100, col = "#0056B9", border = "#0056B9", bty = "n" )
    
 
 
    
 
@@ -1094,7 +1120,7 @@ 
    SVAR-t: Degrees-of-freedom
 
    bsvars modeling of monetary policy
    
 
    SVAR-t: Fitted Values.
    
 
    
-
    post |> compute_fitted_values() |> plot(col = "#0056B9")
    
+
    post |> compute_fitted_values() |> plot(col = "#0056B9")
    
 
 
    
 
diff --git a/index.html b/index.html
index f47f63d..f916cc3 100644
--- a/index.html
+++ b/index.html
@@ -536,6 +536,23 @@ 
    b
 
    - progress bar
    
 
    
 
+
    
+
    My first steps with bsvars
    
+
    Exercise.
    
+
      
+
    1. Download R script w4UKR_bsvars.R
      2. 
+
    2. Open the file in your RStudio
      3. 
+
    3. Execute the code line by line
      4. 
+
    
+
    You have just estimated your first model using the bsvars package! YAY!
    
+
      
+
    1. Modify the line specifying the model to
      2. 
+
    
+
    specify_bsvar_t$new(p = 4) |>
    
+
      
+
    1. Run the code line by line
      2. 
+
    
+
    
 
    
 
    bsvars models and identification
    
 
    
@@ -829,6 +846,15 @@ 
    …for autoregressive parameters.
    
 
  • use the verify_autoregression() function
    • 
 
 
+
    
+
    bsvars hypothesis verification
    
+
    …for autoregressive parameters.
    
+
    Exercise.
    
+
      
+
    1. Download the R script w4UKR_bsvars_Granger.R
      2. 
+
    2. Verify the hypothesis of no Granger causality from ttr to gdp
      3. 
+
    
+
    
 
    
 
    bsvars modeling of monetary policy
    
 
    
@@ -903,36 +929,36 @@ 
    Identified shocks.
    
 
    bsvars modeling of monetary policy
    
 
    Specify and estimate a lower-triangular SVAR-SV.
    
 
    
-
    library(bsvars)
-load("soe.rda")
-soe           = as.matrix(soe)
-
-TT            = nrow(soe)
-lag_order     = 8
-lag_exogenous = 4
-T             = TT - max(lag_order, lag_exogenous)
-
-exogenous     = matrix(NA, TT - lag_exogenous, 0)
-for (i in 0:lag_exogenous) {
-  exogenous   = cbind(exogenous, as.matrix(soe[(lag_exogenous - i + 1):(TT - i), 6:8]))
-}
-
-set.seed(1234)
-spec          = specify_bsvar_sv$new(
-  data        = tail(soe[,1:5], T),
-  p           = lag_order,
-  exogenous   = tail(exogenous, T)
-)
-
-burn          = estimate(spec, 1e4)
-post          = estimate(burn, 1e4)
    
+
    library(bsvars)
+load("soe.rda")
+soe           = as.matrix(soe)
+
+TT            = nrow(soe)
+lag_order     = 8
+lag_exogenous = 4
+T             = TT - max(lag_order, lag_exogenous)
+
+exogenous     = matrix(NA, TT - lag_exogenous, 0)
+for (i in 0:lag_exogenous) {
+  exogenous   = cbind(exogenous, as.matrix(soe[(lag_exogenous - i + 1):(TT - i), 6:8]))
+}
+
+set.seed(1234)
+spec          = specify_bsvar_sv$new(
+  data        = tail(soe[,1:5], T),
+  p           = lag_order,
+  exogenous   = tail(exogenous, T)
+)
+
+burn          = estimate(spec, 1e4)
+post          = estimate(burn, 1e4)
    
 
    
 
 
    
 
    bsvars modeling of monetary policy
    
 
    SVAR-SV: Compute and plot impulse responses.
    
 
    
-
    post |> compute_impulse_responses(horizon = 60) |> plot(probability = 0.68, col = "#0056B9")
    
+
    post |> compute_impulse_responses(horizon = 60) |> plot(probability = 0.68, col = "#0056B9")
    
 
 
    
 
    
@@ -940,7 +966,7 @@ 
    SVAR-SV: Compute and plot i
 
    bsvars modeling of monetary policy
    
 
    SVAR-SV: Forecast error variance decompositions.
    
 
    
-
    post |> compute_variance_decompositions(horizon = 60) |> plot(col = bsvars_grad)
    
+
    post |> compute_variance_decompositions(horizon = 60) |> plot(col = bsvars_grad)
    
 
 
    
 
@@ -948,7 +974,7 @@ 
    SVAR-SV: Forecast error
 
    bsvars modeling of monetary policy
    
 
    SVAR-SV: Structural shocks.
    
 
    
-
    post |> compute_structural_shocks() |> plot(col = "#0056B9")
    
+
    post |> compute_structural_shocks() |> plot(col = "#0056B9")
    
 
 
    
 
@@ -956,7 +982,7 @@ 
    SVAR-SV: Structural shocks.
    
 
    bsvars modeling of monetary policy
    
 
    SVAR-SV: Shocks’ conditional standard deviations.
    
 
    
-
    post |> compute_conditional_sd() |> plot(col = "#0056B9")
    
+
    post |> compute_conditional_sd() |> plot(col = "#0056B9")
    
 
 
    
 
@@ -964,7 +990,7 @@ 
    SVAR-SV: Shocks’ cond
 
    bsvars modeling of monetary policy
    
 
    SVAR-SV: Homoskedasticity verification.
    
 
    
-
    post |> verify_identification() |> summary()
    
+
    post |> verify_identification() |> summary()
    
 
    
 
     **************************************************|
  bsvars: Bayesian Structural Vector Autoregressions|
@@ -988,7 +1014,7 @@ 
    SVAR-SV: Homoskedasticity verifi
 
    bsvars modeling of monetary policy
    
 
    SVAR-SV: Fitted Values.
    
 
    
-
    post |> compute_fitted_values() |> plot(col = "#0056B9")
    
+
    post |> compute_fitted_values() |> plot(col = "#0056B9")
    
 
 
    
 
@@ -996,9 +1022,9 @@ 
    SVAR-SV: Fitted Values.
    
 
    bsvars modeling of monetary policy
    
 
    SVAR-SV: Does foreign sector matter?
    
 
    
-
    A0 = matrix(NA, 5, 56)
-A0[,45:56] = 0
-post |> verify_autoregression(hypothesis = A0) |> summary()
    
+
    A0 = matrix(NA, 5, 56)
+A0[,45:56] = 0
+post |> verify_autoregression(hypothesis = A0) |> summary()
    
 
    
 
     **************************************************|
  bsvars: Bayesian Structural Vector Autoregressions|
@@ -1023,22 +1049,22 @@ 
    SVAR-SV: Does foreign sector matter?
 
    bsvars modeling of monetary policy
    
 
    Specify and estimate a lower-triangular SVAR-t.
    
 
    
-
    set.seed(1234)
-spec          = specify_bsvar_t$new(
-  data        = tail(soe[,1:5], T),
-  p           = lag_order,
-  exogenous   = tail(exogenous, T)
-)
-
-burn          = estimate(spec, 1e4)
-post          = estimate(burn, 1e4)
    
+
    set.seed(1234)
+spec          = specify_bsvar_t$new(
+  data        = tail(soe[,1:5], T),
+  p           = lag_order,
+  exogenous   = tail(exogenous, T)
+)
+
+burn          = estimate(spec, 1e4)
+post          = estimate(burn, 1e4)
    
 
    
 
 
    
 
    bsvars modeling of monetary policy
    
 
    SVAR-t: Compute and plot impulse responses.
    
 
    
-
    post |> compute_impulse_responses(horizon = 60) |> plot(probability = 0.68, col = "#0056B9" )
    
+
    post |> compute_impulse_responses(horizon = 60) |> plot(probability = 0.68, col = "#0056B9" )
    
 
 
    
 
    
@@ -1046,7 +1072,7 @@ 
    SVAR-t: Compute and plot imp
 
    bsvars modeling of monetary policy
    
 
    SVAR-t: Forecast error variance decompositions.
    
 
    
-
    post |> compute_variance_decompositions(horizon = 60) |> plot(col = bsvars_grad )
    
+
    post |> compute_variance_decompositions(horizon = 60) |> plot(col = bsvars_grad )
    
 
 
    
 
@@ -1054,7 +1080,7 @@ 
    SVAR-t: Forecast error v
 
    bsvars modeling of monetary policy
    
 
    SVAR-t: Structural shocks.
    
 
    
-
    post |> compute_structural_shocks() |> plot(col = "#0056B9" )
    
+
    post |> compute_structural_shocks() |> plot(col = "#0056B9" )
    
 
 
    
 
@@ -1062,7 +1088,7 @@ 
    SVAR-t: Structural shocks.
    
 
    bsvars modeling of monetary policy
    
 
    SVAR-t: Normality verification.
    
 
    
-
    post |> verify_identification() |> summary( )
    
+
    post |> verify_identification() |> summary( )
    
 
    
 
     **************************************************|
  bsvars: Bayesian Structural Vector Autoregressions|
@@ -1086,7 +1112,7 @@ 
    SVAR-t: Normality verification.
    
 
    bsvars modeling of monetary policy
    
 
    SVAR-t: Degrees-of-freedom posterior density.
    
 
    
-
    post$posterior$df |> hist(breaks = 100, col = "#0056B9", border = "#0056B9", bty = "n" )
    
+
    post$posterior$df |> hist(breaks = 100, col = "#0056B9", border = "#0056B9", bty = "n" )
    
 
 
    
 
@@ -1094,7 +1120,7 @@ 
    SVAR-t: Degrees-of-freedom
 
    bsvars modeling of monetary policy
    
 
    SVAR-t: Fitted Values.
    
 
    
-
    post |> compute_fitted_values() |> plot(col = "#0056B9")
    
+
    post |> compute_fitted_values() |> plot(col = "#0056B9")
    
 
 
    
 
diff --git a/index.qmd b/index.qmd
index 3ca2f0e..30ef3fd 100644
--- a/index.qmd
+++ b/index.qmd
@@ -259,6 +259,21 @@ spec2 |>
 
 
 
+## My first steps with [bsvars](https://cran.r-project.org/package=bsvars){style="color:#FFD800;"}
+
+### Exercise.
+
+1. Download **R** script [w4UKR_bsvars.R](https://github.com/bsvars/2024-08-bsvars-w4UKR/blob/main/w4UKR_bsvars.R)
+2. Open the file in your **RStudio**
+3. Execute the code line by line
+
+#### You have just estimated your first model using the **bsvars** package! YAY!
+
+4. Modify the line specifying the model to 
+```
+specify_bsvar_t$new(p = 4) |>
+```
+5. Run the code line by line
 
 
 
@@ -736,6 +751,14 @@ $$ $$
 
 
 
+## [bsvars](https://cran.r-project.org/package=bsvars){style="color:#FFD800;"} hypothesis verification
+
+### ...for autoregressive parameters.
+
+### Exercise.
+
+1. Download the **R** script [w4UKR_bsvars_Granger.R](https://github.com/bsvars/2024-08-bsvars-w4UKR/blob/main/w4UKR_bsvars_Granger.R)
+2. Verify the hypothesis of no Granger causality from ttr to gdp
 
 
 ## [bsvars](https://cran.r-project.org/package=bsvars){style="color:#FFD800;"} modeling of monetary policy   {background-color="#0056B9"}
diff --git a/w4UKR_bsvars_Granger.R b/w4UKR_bsvars_Granger.R
index e69de29..4a8c1b0 100644
--- a/w4UKR_bsvars_Granger.R
+++ b/w4UKR_bsvars_Granger.R
@@ -0,0 +1,19 @@
+
+
+library(bsvars)
+
+# upload data
+data(us_fiscal_lsuw)    # upload dependent variables
+
+set.seed(123)
+us_fiscal_lsuw |>
+  specify_bsvar$new() |>
+  estimate(S = 1000) |> 
+  estimate(S = 2000) -> post
+
+post$last_draw$starting_values$A # study the structure of the A matrix
+
+# verify no Granger causality from ttr to gdp
+H0      = matrix(NA, 3, 4)
+H0[,]    = 0 # modify this line to set appropriate restriction
+post |> verify_autoregression(hypothesis = H0) |> summary()