Load necessary packages library(urca) library(vars) library(tseries) library(MASS)…
- User
Load necessary packages
library(urca) library(vars) library(tseries) library(MASS) library(tidyverse)
Assuming Y is already loaded in the environment
Select the relevant columns
Y <- log_data[, c("log_Spot", "log_Fut", "log_USO", "log_USL", "log_OIL")]
View structure of Y
str(Y) tibble [3,031 × 5] (S3: tbl_df/tbl/data.frame) log_Fut : num [1:3031] 4.12 4.12 4.12 4.11 4.1 ... log_USL : num [1:3031] 3.14 3.15 3.14 3.14 3.13 ... $ log_OIL : num [1:3031] 5.64 5.64 5.64 5.63 5.62 ...
- attr(, "na.action")= 'omit' Named int [1:6847] 1 2 3 4 5 6 7 8 9 10 ... ..- attr(, "names")= chr [1:6847] "1" "2" "3" "4" ...
Step 1: Unit root tests using Phillips-Perron (PP) test and KPSS test
Initialize data frames to store test results
pp_results <- data.frame(Series=character(), Test_Statistic=numeric(), p_value=numeric(), stringsAsFactors=FALSE) kpss_results <- data.frame(Series=character(), Test_Statistic=numeric(), p_value=numeric(), stringsAsFactors=FALSE)
Loop over each column in Y
for(series in names(Y)) {
-
Phillips-Perron test
- pp_test <- pp.test(Y[[series]])
- pp_results <- rbind(pp_results, data.frame(Series=series, Test_Statistic=pp_testp.value))
-
KPSS test
- kpss_test <- kpss.test(Y[[series]], null="Level")
- kpss_results <- rbind(kpss_results, data.frame(Series=series, Test_Statistic=kpss_testp.value))
- } Warning messages: 1: In kpss.test(Y[[series]], null = "Level") : p-value smaller than printed p-value 2: In kpss.test(Y[[series]], null = "Level") : p-value smaller than printed p-value 3: In kpss.test(Y[[series]], null = "Level") : p-value smaller than printed p-value 4: In kpss.test(Y[[series]], null = "Level") : p-value smaller than printed p-value 5: In kpss.test(Y[[series]], null = "Level") : p-value smaller than printed p-value
Print the results
print("Phillips-Perron Test Results:") [1] "Phillips-Perron Test Results:" print(pp_results) Series Test_Statistic p_value Dickey-Fuller Z(alpha) log_Spot -10.237353 0.5387773 Dickey-Fuller Z(alpha)1 log_Fut -9.894884 0.5578765 Dickey-Fuller Z(alpha)2 log_USO -8.401415 0.6411661 Dickey-Fuller Z(alpha)3 log_USL -9.306146 0.5907100 Dickey-Fuller Z(alpha)4 log_OIL -7.929339 0.6674934 print("KPSS Test Results:") [1] "KPSS Test Results:" print(kpss_results) Series Test_Statistic p_value KPSS Level log_Spot 10.48372 0.01 KPSS Level1 log_Fut 10.61626 0.01 KPSS Level2 log_USO 24.56629 0.01 KPSS Level3 log_USL 20.71650 0.01 KPSS Level4 log_OIL 24.82409 0.01
Test first differences
Initialize data frames to store test results for first differences
pp_results_diff <- data.frame(Series=character(), Test_Statistic=numeric(), p_value=numeric(), stringsAsFactors=FALSE) kpss_results_diff <- data.frame(Series=character(), Test_Statistic=numeric(), p_value=numeric(), stringsAsFactors=FALSE)
Loop over each column in Y
for(series in names(Y)) {
- diff_series <- diff(Y[[series]])
-
Phillips-Perron test
- pp_test <- pp.test(diff_series)
- pp_results_diff <- rbind(pp_results_diff, data.frame(Series=series, Test_Statistic=pp_testp.value))
-
KPSS test
- kpss_test <- kpss.test(diff_series, null="Level")
- kpss_results_diff <- rbind(kpss_results_diff, data.frame(Series=series, Test_Statistic=kpss_testp.value))
- } Warning messages: 1: In pp.test(diff_series) : p-value smaller than printed p-value 2: In kpss.test(diff_series, null = "Level") : p-value greater than printed p-value 3: In pp.test(diff_series) : p-value smaller than printed p-value 4: In kpss.test(diff_series, null = "Level") : p-value greater than printed p-value 5: In pp.test(diff_series) : p-value smaller than printed p-value 6: In kpss.test(diff_series, null = "Level") : p-value greater than printed p-value 7: In pp.test(diff_series) : p-value smaller than printed p-value 8: In kpss.test(diff_series, null = "Level") : p-value greater than printed p-value 9: In pp.test(diff_series) : p-value smaller than printed p-value 10: In kpss.test(diff_series, null = "Level") : p-value greater than printed p-value
Print the results for first differences
print("Phillips-Perron Test Results for First Differences:") [1] "Phillips-Perron Test Results for First Differences:" print(pp_results_diff) Series Test_Statistic p_value Dickey-Fuller Z(alpha) log_Spot -3041.242 0.01 Dickey-Fuller Z(alpha)1 log_Fut -3164.641 0.01 Dickey-Fuller Z(alpha)2 log_USO -3217.069 0.01 Dickey-Fuller Z(alpha)3 log_USL -3208.247 0.01 Dickey-Fuller Z(alpha)4 log_OIL -3212.187 0.01 print("KPSS Test Results for First Differences:") [1] "KPSS Test Results for First Differences:" print(kpss_results_diff) Series Test_Statistic p_value KPSS Level log_Spot 0.05516311 0.1 KPSS Level1 log_Fut 0.05715371 0.1 KPSS Level2 log_USO 0.10059645 0.1 KPSS Level3 log_USL 0.07285208 0.1 KPSS Level4 log_OIL 0.10775410 0.1
Step 2: Determine the optimal lag length using VARselect
VARselect_result <- VARselect(Y, lag.max=10, type="const") print("Optimal lag length selection:") [1] "Optimal lag length selection:" print(VARselect_result$selection) AIC(n) HQ(n) SC(n) FPE(n) 10 9 5 10
Choose the number of lags based on AIC or other criteria
We'll use the lag order suggested by AIC
p <- VARselect_result$selection["SC(n)"]
Step 3: Johansen cointegration test
Since the Johansen test in urca requires the number of lags in levels,
we use K = p, where p is the number of lags in the VAR
johansen_test <- ca.jo(Y, type="trace", ecdet="const", K=p, spec="transitory") summary(johansen_test)
######################
Johansen-Procedure
######################
Test type: trace statistic , without linear trend and constant in cointegration
Eigenvalues (lambda): [1] 1.376445e-01 5.485084e-02 1.460022e-02 6.534633e-03 7.806915e-04 2.873217e-17
Values of teststatistic and critical values of test:
test 10pct 5pct 1pctr <= 4 | 2.36 7.52 9.24 12.97 r <= 3 | 22.20 17.85 19.96 24.60 r <= 2 | 66.71 32.00 34.91 41.07 r <= 1 | 237.41 49.65 53.12 60.16 r = 0 | 685.53 71.86 76.07 84.45
Eigenvectors, normalised to first column: (These are the cointegration relations)
log_Spot.l1 log_Fut.l1 log_USO.l1 log_USL.l1 log_OIL.l1 constantlog_Spot.l1 1.0000000000 1.0000000 1.00000 1.0000000 1.000000 1.000000 log_Fut.l1 -1.0017520247 -1.3244152 16.80517 0.7523886 -8.329693 -1.820012 log_USO.l1 0.0023907033 -0.5014035 -74.82706 8.6459957 15.122284 -9.148170 log_USL.l1 -0.0043867524 0.5028687 -35.29678 -6.3180603 -4.415480 5.683194 log_OIL.l1 0.0005538807 0.2823013 76.35024 -4.8355656 -8.218566 3.311713 constant 0.0069055220 0.5034032 -47.90226 -0.2647043 16.978031 10.890075
Weights W: (This is the loading matrix)
log_Spot.l1 log_Fut.l1 log_USO.l1 log_USL.l1 log_OIL.l1 constantlog_Spot.d -0.24199295 0.02295218 0.0003860446 -0.0045865854 0.0001805646 -3.849043e-15 log_Fut.d 0.18580099 0.04212123 0.0004176988 -0.0042320772 0.0001764496 1.940801e-15 log_USO.d -0.15331674 0.06760043 0.0005287744 -0.0024737038 0.0001574821 -2.212215e-15 log_USL.d -0.07075644 0.02880841 0.0006447661 -0.0007056863 0.0001551454 -1.849213e-15 log_OIL.d -0.18973739 0.07470091 0.0002518853 -0.0017978899 0.0001975052 -1.958137e-15
Determine the number of cointegrating vectors (r)
From the summary, identify the rank r where the test statistic
is less than the critical value. Suppose r = 4
r = 4
Step 4: Estimate VECM and transform to VAR representation
vecm_var <- vec2var(johansen_test, r=r) print(vecm_var)
Coefficient matrix of lagged endogenous variables:
A1: log_Spot.l1 log_Fut.l1 log_USO.l1 log_USL.l1 log_OIL.l1 log_Spot 1.0314300 -0.11368162 -0.44715044 0.1682500 0.3250480 log_Fut 0.2645245 0.59622236 -0.42963580 0.1877418 0.3369964 log_USO 0.2434639 0.05889863 0.03556935 0.3508134 0.2977061 log_USL 0.2032903 -0.02635639 -0.42589721 1.0087496 0.1832933 log_OIL 0.2104761 0.08089268 -0.91049287 0.3401844 1.2535131
A2: log_Spot.l2 log_Fut.l2 log_USO.l2 log_USL.l2 log_OIL.l2 log_Spot -0.25695033 0.17157095 0.3068445 0.04945633 -0.1870758 log_Fut -0.04085083 -0.04446826 0.3808412 -0.03100964 -0.1976518 log_USO -0.40256986 0.06270538 0.6244124 -0.07113705 -0.1625828 log_USL -0.34630385 0.08910242 0.4026364 0.05302569 -0.1230994 log_OIL -0.37715993 0.04294832 0.7207870 -0.06653196 -0.2329700
A3: log_Spot.l3 log_Fut.l3 log_USO.l3 log_USL.l3 log_OIL.l3 log_Spot 0.1458521 -0.1510641 0.3339362 -0.3276682 -0.027737351 log_Fut 0.2034098 -0.1089163 0.1496027 -0.2465002 -0.015756806 log_USO 0.1856833 -0.2871774 0.2521153 -0.1912245 0.005717099 log_USL 0.1226135 -0.2350356 0.1789151 -0.1471299 0.038250937 log_OIL 0.2326656 -0.3121597 0.2278128 -0.1744500 -0.025562538
A4: log_Spot.l4 log_Fut.l4 log_USO.l4 log_USL.l4 log_OIL.l4 log_Spot -0.18863718 0.1710641 0.19081976 0.01677350 -0.1791175 log_Fut -0.25618937 0.1942828 0.20075854 0.09335346 -0.1992789 log_USO -0.18182951 0.2347955 0.16972329 -0.03358236 -0.1772673 log_USL -0.04536394 0.1022300 0.09913327 0.04443748 -0.1707412 log_OIL -0.31741773 0.3324992 0.18859322 -0.04652779 -0.1442159
A5: log_Spot.l5 log_Fut.l5 log_USO.l5 log_USL.l5 log_OIL.l5 log_Spot 0.04506413 0.137166040 -0.4650790 0.12114404 0.1268813 log_Fut 0.05321376 0.124802235 -0.3900878 0.02877604 0.1400407 log_USO 0.06759086 0.001857149 -0.1770362 -0.02323782 0.1077595 log_USL 0.02375506 0.113090122 -0.3237487 0.03741476 0.1330303 log_OIL 0.13485349 -0.050165498 -0.2990014 -0.01180909 0.1981438
Coefficient matrix of deterministic regressor(s).
constantlog_Spot -0.007395204 log_Fut 0.003598548 log_USO 0.008296858 log_USL -0.016685317 log_OIL 0.024704477
Step 5: Extract beta and alpha matrices
beta <- johansen_test@V[1:r+1, 1:r] alpha <- johansen_test@W[, 1:r] print("Extracted beta matrix:") [1] "Extracted beta matrix:" print(beta) log_Spot.l1 log_Fut.l1 log_USO.l1 log_USL.l1 log_Fut.l1 -1.0017520247 -1.3244152 16.80517 0.7523886 log_USO.l1 0.0023907033 -0.5014035 -74.82706 8.6459957 log_USL.l1 -0.0043867524 0.5028687 -35.29678 -6.3180603 log_OIL.l1 0.0005538807 0.2823013 76.35024 -4.8355656 print("Extracted alpha matrix:") [1] "Extracted alpha matrix:" print(alpha) log_Spot.l1 log_Fut.l1 log_USO.l1 log_USL.l1 log_Spot.d -0.24199295 0.02295218 0.0003860446 -0.0045865854 log_Fut.d 0.18580099 0.04212123 0.0004176988 -0.0042320772 log_USO.d -0.15331674 0.06760043 0.0005287744 -0.0024737038 log_USL.d -0.07075644 0.02880841 0.0006447661 -0.0007056863 log_OIL.d -0.18973739 0.07470091 0.0002518853 -0.0017978899
Step 6: Compute orthogonal complements beta_perp and alpha_perp
Use Null space computation from MASS package
beta_perp <- Null(t(beta)) alpha_perp <- Null(t(alpha)) print("Computed orthogonal complement beta_perp:") [1] "Computed orthogonal complement beta_perp:" print(beta_perp)
[1,] [2,] [3,] [4,]
print("Computed orthogonal complement alpha_perp:") [1] "Computed orthogonal complement alpha_perp:" print(alpha_perp)
[1,] [2,] [3,] [4,]
Ensure beta_perp and alpha_perp are matrices
beta_perp <- as.matrix(beta_perp) alpha_perp <- as.matrix(alpha_perp) print("Converted beta_perp to matrix:") [1] "Converted beta_perp to matrix:" print(beta_perp)
[1,] [2,] [3,] [4,]
print("Converted alpha_perp to matrix:") [1] "Converted alpha_perp to matrix:" print(alpha_perp)
[1,] [2,] [3,] [4,]
Step 7: Compute long-run impact matrix Psi(1)
Psi_1 <- beta_perp %*% t(alpha_perp) print("Computed long-run impact matrix Psi(1):") [1] "Computed long-run impact matrix Psi(1):" print(Psi_1) [,1] [,2] [,3] [,4] [1,] 0 0 0 0 [2,] 0 0 0 0 [3,] 0 0 0 0 [4,] 0 0 0 0
Step 8: Compute residual covariance matrix Omega
residuals_var <- residuals(vecm_var) Omega <- cov(residuals_var) print("Computed residual covariance matrix Omega:") [1] "Computed residual covariance matrix Omega:" print(Omega) resids of log_Spot resids of log_Fut resids of log_USO resids of log_USL resids of log_OIL resids of log_Spot 0.0005644515 0.0005321534 0.0004577378 0.0003636763 0.0004834179 resids of log_Fut 0.0005321534 0.0005445740 0.0004583258 0.0003635811 0.0004826371 resids of log_USO 0.0004577378 0.0004583258 0.0004424266 0.0003554061 0.0004709637 resids of log_USL 0.0003636763 0.0003635811 0.0003554061 0.0003310452 0.0003825727 resids of log_OIL 0.0004834179 0.0004826371 0.0004709637 0.0003825727 0.0005346706
Step 9: Compute residual correlation matrix and its eigenvalues and eigenvectors
corr_matrix <- cov2cor(Omega) eigen_result <- eigen(corr_matrix) Lambda <- diag(eigen_resultvectors print("Computed residual correlation matrix:") [1] "Computed residual correlation matrix:" print(corr_matrix) resids of log_Spot resids of log_Fut resids of log_USO resids of log_USL resids of log_OIL resids of log_Spot 1.0000000 0.9598318 0.9159733 0.8413140 0.8799668 resids of log_Fut 0.9598318 1.0000000 0.9337385 0.8563065 0.8944358 resids of log_USO 0.9159733 0.9337385 1.0000000 0.9286681 0.9683306 resids of log_USL 0.8413140 0.8563065 0.9286681 1.0000000 0.9093419 resids of log_OIL 0.8799668 0.8944358 0.9683306 0.9093419 1.0000000 print("Eigenvalues of correlation matrix:") [1] "Eigenvalues of correlation matrix:" print(eigen_result$values) [1] 4.63622032 0.21050792 0.09032045 0.04009674 0.02285457 print("Eigenvectors of correlation matrix:") [1] "Eigenvectors of correlation matrix:" print(G) [,1] [,2] [,3] [,4] [,5] [1,] -0.4434738 0.5646226 -0.18415788 0.66946900 -0.04928965 [2,] -0.4480840 0.4715324 -0.09924278 -0.70169019 0.27323925 [3,] -0.4580114 -0.1500695 0.26353846 -0.16465350 -0.81923222 [4,] -0.4373571 -0.5864481 -0.66839993 0.03065071 0.13076471 [5,] -0.4488832 -0.3040004 0.66334437 0.17717761 0.48442751
Step 10: Compute Cholesky decomposition of Omega
V <- t(chol(Omega)) print("Computed Cholesky decomposition of Omega (matrix V):") [1] "Computed Cholesky decomposition of Omega (matrix V):" print(V) resids of log_Spot resids of log_Fut resids of log_USO resids of log_USL resids of log_OIL resids of log_Spot 0.02375819 0.000000000 0.000000000 0.0000000000 0.000000000 resids of log_Fut 0.02239874 0.006547554 0.000000000 0.0000000000 0.000000000 resids of log_USO 0.01926653 0.004090063 0.007382338 0.0000000000 0.000000000 resids of log_USL 0.01530741 0.003163691 0.006440466 0.0067260564 0.000000000 resids of log_OIL 0.02034742 0.004105427 0.008418477 0.0005797034 0.005708903
Step 11: Compute F^M matrix
Lambda_inv_sqrt <- diag(1 / sqrt(eigen_result$values)) temp_matrix <- G %% Lambda_inv_sqrt %% t(G) V_inv <- solve(V) F_M_inv <- temp_matrix %*% V_inv F_M <- solve(F_M_inv) print("Computed F^M inverse matrix:") [1] "Computed F^M inverse matrix:" print(F_M_inv) resids of log_Spot resids of log_Fut resids of log_USO resids of log_USL resids of log_OIL [1,] 394.28409 -219.65122 -4.123141 -20.28489 -44.44236 [2,] -552.85929 628.98670 -86.952894 -21.32204 -32.12105 [3,] -16.70373 -616.82561 1177.184073 -123.78615 -349.31488 [4,] 50.63079 77.57887 -354.546552 371.47613 -96.23931 [5,] 67.67344 184.29224 -845.963463 -134.04522 607.50859 print("Computed F^M matrix:") [1] "Computed F^M matrix:" print(F_M) [,1] [,2] [,3] [,4] [,5] resids of log_Spot 0.01591893 0.01091479 0.008741038 0.007265960 0.007918760 resids of log_Fut 0.01801606 0.01452761 0.010800059 0.008920024 0.009709158 resids of log_USO 0.01750445 0.01438370 0.013041080 0.010149610 0.011147485 resids of log_USL 0.01613659 0.01372343 0.013367719 0.013021034 0.011655206 resids of log_OIL 0.02069707 0.01743461 0.016859437 0.013491171 0.015913318
Step 12: Compute psi_1^G
psi_1_r <- Psi_1[1, ] # First row of Psi(1) psi_1_G <- psi_1_r %% F_M Error in psi_1_r %% F_M : non-conformable arguments
- Assistant
It seems like you're encountering an error due to a dimension mismatch during the matrix multiplication for computing
psi_1^G. Let's walk through this step carefully, ensuring all matrix dimensions align properly. Here's a brief breakdown of the steps leading to this point and adjustments you might need:-
Psi(1) Matrix: You've computed
Psi_1as a zeros matrix, which may reflect an earlier computational step that didn't yield the expected non-zero values. Verify ifPsi_1is supposed to be non-zero based on your understanding of the model or data. -
Dimension Compatibility: When you attempt to compute
psi_1^Gby multiplyingpsi_1_rwithF_M, ensure the dimensions are compatible. Typically,psi_1_rshould be a row vector andF_Ma matrix where the number of columns inpsi_1_rmatches the number of rows inF_M. -
Checking Dimensions: Use the
dim()function in R to check the dimensions of matrices or vectors involved before performing matrix multiplication:print(dim(psi_1_r)) # Should output the dimensions of psi_1_r print(dim(F_M)) # Should output the dimensions of F_M -
Adjusting Dimensions: Ensure that either
psi_1_rorF_M(or both) are transposed if needed to make the number of columns inpsi_1_requal to the number of rows inF_M. -
Correct the Calculation: If
psi_1_rremains a row of zeros, re-examine howPsi_1was computed. It seemsbeta_perpandalpha_perpare zero matrices, possibly suggesting the computation of null spaces didn't return expected results. Verify ifbetaoralpharequire a different selection of cointegration rank or if calculation of their null spaces needs adjustment.
Here’s a conceptual snippet adjusted for proper dimension matching:
# Check if Psi_1 and F_M have compatible dimensions for multiplication if (all(dim(psi_1_r)) != c(1, nrow(F_M))) { stop("Dimensions not aligned, adjust inputs for matrix multiplication.") } # Corrected multiplication assuming psi_1_r's and F_M's dimensions match appropriately psi_1_G <- psi_1_r %*% F_M print("Computed psi_1^G:") print(psi_1_G)If
psi_1_rremains zero, go back and revisit the steps for calculatingbeta_perpandalpha_perpto ensure the computation of null spaces properly reflects your model specifications and produces expected values. -