0
$\begingroup$

I'm struggling with the interpretation of one of my course examples.

Here is the R output

 library(urca)
    rates.co <- ca.jo(data_rate, type = "eigen", ecdet = "const", K = 2, spec = 'transitory', season = NULL) 

    summary(rates.co)
## 
## ###################### 
## # Johansen-Procedure # 
## ###################### 
## 
## Test type: maximal eigenvalue statistic (lambda max) , without linear trend and constant in cointegration 
## 
## Eigenvalues (lambda):
## [1]  2.839734e-01  1.863156e-01  1.390916e-01  7.538804e-02  3.502591e-02
## [6] -5.514428e-17
## 
## Values of teststatistic and critical values of test:
## 
##           test 10pct  5pct  1pct
## r <= 4 |  7.56  7.52  9.24 12.97
## r <= 3 | 16.62 13.75 15.67 20.20
## r <= 2 | 31.75 19.77 22.00 26.81
## r <= 1 | 43.71 25.56 28.14 33.24
## r = 0  | 70.82 31.66 34.40 39.79
## 
## Eigenvectors, normalised to first column:
## (These are the cointegration relations)
## 
##                cofi.l1  X1ycmt.l1  X5ycmt.l1 primeRate.l1 X3mTbill.l1
## cofi.l1       1.000000  1.0000000  1.0000000     1.000000  1.00000000
## X1ycmt.l1    -5.016228  2.8868596  2.3696681     1.613084  2.44060485
## X5ycmt.l1     1.447589 -0.9385606 -1.3010429    -1.864387 -3.90107940
## primeRate.l1  1.798685  0.2357253  0.9647242    -3.533503 -2.05938244
## X3mTbill.l1   1.274463 -3.3546732 -3.1195849     2.575369 -0.04422163
## constant     -7.582615 -0.2506138 -2.4391828    13.674437 20.63481751
##                constant
## cofi.l1       1.0000000
## X1ycmt.l1    -0.9501416
## X5ycmt.l1     0.6813733
## primeRate.l1 -2.7068548
## X3mTbill.l1   2.2016130
## constant      3.7872633
## 


## Weights W:
## (This is the loading matrix)
## 
##                 cofi.l1    X1ycmt.l1    X5ycmt.l1 primeRate.l1
## cofi.d      -0.03002766 -0.025996453  0.001132855 -0.005679920
## X1ycmt.d    -0.02895731  0.034635064 -0.021423063 -0.005731646
## X5ycmt.d    -0.02949078  0.000379748  0.041174443  0.018020526
## primeRate.d -0.04390107  0.028939965 -0.066368855  0.007284998
## X3mTbill.d  -0.04602132  0.113886433 -0.017776290 -0.011264136
##               X3mTbill.l1      constant
## cofi.d      -2.078558e-05  4.181518e-17
## X1ycmt.d     9.894523e-03  1.014418e-16
## X5ycmt.d     1.145797e-02 -4.228840e-16
## primeRate.d  5.813215e-04  1.985921e-16
## X3mTbill.d   5.261185e-03  1.341192e-16


    library(tsDyn)
    vecm<-VECM(data_rate, lag=2-1, include = 'const', estim = 'ML', r=4) 
    summary(vecm)


## #############
## ###Model VECM 
## #############
## Full sample size: 214    End sample size: 212
## Number of variables: 5   Number of estimated slope parameters 50
## AIC -4907.008    BIC -4725.753   SSR 26.22101
## Cointegrating vector (estimated by ML):
##             cofi         1ycmt         5ycmt     primeRate    3mTbill
## r1  1.000000e+00 -8.326673e-17  0.000000e+00  1.110223e-16 -0.9659268
## r2 -2.081668e-16  1.000000e+00 -1.110223e-16 -5.551115e-17 -1.0839784
## r3 -2.775558e-17  1.110223e-16  1.000000e+00  0.000000e+00 -1.0269771
## r4  5.551115e-17  0.000000e+00  5.551115e-17  1.000000e+00 -0.9514607
## 
## 
##                    ECT1                ECT2              
## Equation cofi      -0.0606(0.0138)***  0.0690(0.0404).   
## Equation 1ycmt     -0.0198(0.0517)     0.1848(0.1518)    
## Equation 5ycmt     0.0320(0.0623)      0.2752(0.1830)    
## Equation primeRate -0.0738(0.0295)*    0.1585(0.0867).   
## Equation 3mTbill   0.0397(0.0405)      0.4991(0.1190)*** 
##                    ECT3                ECT4               
## Equation cofi      -0.0097(0.0160)     -0.0389(0.0162)*   
## Equation 1ycmt     -0.0362(0.0604)     -0.0501(0.0610)    
## Equation 5ycmt     -0.1307(0.0727).    -0.0836(0.0735)    
## Equation primeRate -0.0184(0.0345)     -0.1624(0.0348)*** 
## Equation 3mTbill   -0.1295(0.0473)**   -0.0363(0.0478)    
##                    Intercept          cofi -1            
## Equation cofi      0.1527(0.0593)*    0.2211(0.0647)***  
## Equation 1ycmt     0.2005(0.2230)     -0.0020(0.2434)    
## Equation 5ycmt     0.3884(0.2687)     -0.2470(0.2933)    
## Equation primeRate 0.5904(0.1273)***  0.1548(0.1390)     
## Equation 3mTbill   0.2179(0.1747)     0.1439(0.1907)     
##                    1ycmt -1            5ycmt -1          
## Equation cofi      -0.0422(0.0569)     0.0006(0.0319)    
## Equation 1ycmt     -0.0086(0.2141)     0.2610(0.1200)*   
## Equation 5ycmt     0.1253(0.2580)      0.3309(0.1447)*   
## Equation primeRate -0.0946(0.1223)     0.0696(0.0686)    
## Equation 3mTbill   -0.1980(0.1678)     0.1649(0.0941).   
##                    primeRate -1        3mTbill -1         
## Equation cofi      0.0630(0.0329).     0.0856(0.0494).    
## Equation 1ycmt     -0.0642(0.1237)     0.1667(0.1857)     
## Equation 5ycmt     -0.0211(0.1491)     -0.2666(0.2239)    
## Equation primeRate -0.0156(0.0707)     0.3743(0.1061)***  
## Equation 3mTbill   -0.0234(0.0970)     0.3131(0.1456)*

    round(vecm$model.specific$coint, 3)

##               r1     r2     r3     r4
## cofi       1.000  0.000  0.000  0.000
## 1ycmt      0.000  1.000  0.000  0.000
## 5ycmt      0.000  0.000  1.000  0.000
## primeRate  0.000  0.000  0.000  1.000
## 3mTbill   -0.966 -1.084 -1.027 -0.951

So from what I understood, I got 4 cointegrating vectors since my test rejected $r \leq 3$ and accepted $r \leq 4$ at 5% critical level.

At the very end I have 4 cointegrating vectors written in a column, which, from what I understood, form a matrix for $\beta$ in the VECM.

What I do not understand though is how to derive the $\alpha$ matrix from this output or, generally speaking, how do I write my final equation for the ECM in a matrix form.

|cite|improve this question
$\endgroup$
1
$\begingroup$

The first r columns of the loading matrix gives you $\alpha$.

You can also get all the coefficients for the VECM easily with the cajorls() function:

vecm <- cajorls(rates.co)
vecm$rlm$coefficients 
vecm$beta

The VECM in matrix form is: $\Delta X_{t} = \alpha(\beta' X_{t-1} - \mu) + \Gamma_{1} \Delta X_{t-1} + \epsilon_{t}$ where $\alpha$ is the first four columns of the loading matrix and indicates how quickly the rates converge to the equilibrium relationship. $\beta$ contains the weights on the rates in the co-integrating relationships while the constant terms for the relationships go in $\mu$. $\Gamma_{1}$ is not show in the ca.jo summary output but can be shown with ca.jorls().

|cite|improve this answer
$\endgroup$
  • $\begingroup$ Welcome to our site. This question remains because it has a statistical component: "how do I write my final equation for the ECM in a matrix form?" Would you have an answer to that? $\endgroup$ – whuber♦ Feb 13 at 18:31
  • $\begingroup$ The ECM in matrix form is: $\Delta X_{t} = \alpha (\beta' X_{t-1} - \mu) + \Gamma_{1} \Delta X_{t-1} + \epsilon_{t}$ where $\alpha$ is the first 4 columns of the loading matrix and indicates how quickly the rates converge to the equilibrium relationship. $\beta$ contains the weights on the rates of the co-integrating relationships with the constant terms in $\mu$. $\Gamma_{1}$ is not shown in the ca.jo output but can be shown with ca.jorls. $\endgroup$ – Matt P Feb 14 at 13:43
  • $\begingroup$ Please edit your post to include your answer. That will make it much more useful and clear. $\endgroup$ – whuber♦ Feb 14 at 15:36
  • $\begingroup$ Thank you for the answer, gladly I figured it out eventually and your answer matches my result $\endgroup$ – Makina Feb 15 at 10:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.