CATS is an add-on program to RATS: Regression Analysis of Time Series , the cointegration facilities in Microfit, and a beta version of PC-FIML 8 is. By David Tufte; CATS in RATS: cointegration analysis of time series: version . CATS in RATS: Cointegration Analysis of Time Series. Front Cover. Henrik Hansen, Katarina Juselius. Estima, – Cointegration – 87 pages.
|Published (Last):||16 November 2013|
|PDF File Size:||20.99 Mb|
|ePub File Size:||20.96 Mb|
|Price:||Free* [*Free Regsitration Required]|
Viewing and Loading Models As noted in the output shown above, the preferred model is loaded globally in CATS enabling you to continue working with it straight away. In this case, you can get CATS to check if the restrictions are generically identifying by anslysis the rank conditions given by 1.
Note that this option is only available when you are analyzing an I. You will then be prompted for the row-wise normalization of.
Convergence Criterion Sets the convergence criterion for the switching algorithm. Under the null, the test is asymptotically distributed as 2. Cointegrating Relations plots the cointegrating relations. Again, we see the effect of the German uni cation on the eigenvalues. Test for Weak Exogeneity imposes zero-rows on the matrix, see paragraph 1. Almost all output generated by RATS instructions, such as tables of regression coefficients, can now be reloaded in a report window via the Report Windows menu operation.
The syntax is quite similar to that used for CATS version 1. Use the User Settings dialog to set the preferred output format. Parameter Constancy Figure 4. Testing Identifying Restrictions on The relations given by 0 X t are at most integrated of order 1 and cointegrate with the differences 1X t to stationarity. Here you specify the discretization of the unit interval T and number of repli- cations N for simulation of the rank test statistic for the I. Setting r D 2, thereby imposing three unit roots, we note that there seem to be no more unit roots, suggesting that the rank of 5 is, indeed, 2.
For the X-form, we see clear rejection in the early recursion, and for the period Then, under the null, the condi- tional model for 1Z t given 1Wt can be reformulated as 0 1Z t D! Test for Unit Vectors in Alpha restricts as D. This is the default setting. In these cases, it is a good idea to look at the graphical behavior of the estimated cointegration rela- tions before deciding about the choice of r.
As discussed in section 1. Export RATS model takes the estimated coef cients of the last model and cointegrayion these to set up a model of RATS equations in either levels or differences as described in section 6. Lag Length Determination calculates tests of lag reduction for the I.
Test for Unit Vectors in Alpha to open the menu in gure 3.
The editor also includes menu-driven Wizards for many tasks, including reading in data, running regressions, doing hypothesis tests, graphing results, and more. To illustrate, we will test if contains e2i. Then all accepted models contained in C1 ;: Dt cointegrafion also contain stationary stochastic variables that are weakly exogenous, or that can be excluded from the cointegrating space.
To save space we only show the top model: The condition for Wt to be weakly exogenous for in 1. Save Restrictions To save the currently imposed restrictions, simply choose I I 1: Sets the convergence criterion for the -switching algorithm. Backwards Recursive Estimation executes the backwards recursive estimation procedure described in sec- tion 4.
This implies that H. To impose the German in ation relation on the rst vector, we choose to impose three restrictions and input the design matrix H1 shown in gure 3.
In this situation it can be useful to reduce the number of basic relations in order to reduce the number of suggested structures as well as computation time. Or, you can check the corresponding p-value reported in brackets below each test statistic.
The editor also offers more than 40 menu-driven Wizards that provide point-and-click access to most common tasks, including reading data, displaying graphs, doing transformations, estimating a variety of models, and hypothesis testing. The elements of C2 are sorted in descending order by p-value. The response on a shock to the German bond rate is higher prices in Germany and a small, probably insigni cant, effect on the US prices.
Maximal Number of Iterations This is the maximal number of iterations performed by the switching procedure estimating.
This gives us the following screen output: In serie case, nine models are found and reported to the output window. As mentioned above, the set of basic relations may be large, and it may also contain many almost identical relations which will result in many almost identical structures when you run the model selection procedure. The Multivariate Cointegration Model variate tests have nonstationarity as the null.
Henrik Hansen was involved in the de- and reconstruction of CATS 1, and the program has bene tted greatly from many discussions in this phase of the development.
The graph seems to give support for this assumption. The default is P2 D 0: Redesigned and expanded toolbar icons Many point-and-click Wizards have new features and are easier to use More control over default arrangement of Input and Output windows.
Restrictions on Each Anakysis Vector imposes restrictions on each of the vectors in. Namely when all or some cointegrating vectors are known, and when the same linear restrictions are applied to all cointegrating vectors. Here, we choose only to include the real interest rate parity relation.