EViews 10 SVARS

[Music] Ede users allowed estimation of structural vector autoregression x' as FAR's for over 20 years as far as are used extensively in macroeconomics is a method of analyzing policy decisions one of the key elements behind s5 estimation is the necessary imposition of restrictions of a residual structural matrices these restrictions generally take the form of restrictions on the factorization matrices a and B restrictions on a short-run impulse response matrix s all restrictions on the long-run impulse response matrix F or C or a combination of the above this version to be used only allowed restrictions on a and B or on F Indies 10 broadens the restriction engine by allowing restrictions on any of the four matrices adding linear restrictions and adds a new interface allowing either specification of the restrictions to demonstrate this new feature I'm going to be working with a few examples taken from a textbook econometrics modelling with time series by Martin earn and Harris 2013 the first example is example fourteen point one three in the book and is working with instantaneous one month and three month us coupon yields we have monthly data from 1947 to 1991 we'll first estimate a standard bar with one lag you I can estimate the structural fall by taking on proc estimate structural factorization this brings up the s file options dialog the first page allows us to impose restrictions we have a number of options for doing so the combo box at the top allows us to change the way we enter the restrictions we can choose from one of the predefined restrictions or specify restrictions by text or completely custom following the book we're going to impose shortly restrictions on both the a and B matrices specifically we're going to impose my analogy on the B matrix and diagonal T on the a matrix with the exception of two free parameters in the lower section we'll do this by selecting a recursive factorization predefined setting then modifying it slightly by restricting one of the lower parameters in a to be 0 leaving B is a predefined diagonal matrix will impose no indirect restrictions on s or F clicking okay produces the results the top part of the results describes what we did including the restrictions we placed on the matrices beneath that we have the estimated coefficients the free parameters and the matrices and then we have finally the estimated matrices themselves now that even though we specify the restrictions in terms of a and B EVs reports to calculate coefficients for all four matrices as a second demonstration we'll move on to example fourteen point one four from the same book this example starts with the same data but this time estimates the bar in differences you this time we are going to impose restrictions on the long run F matrix rather than a or B I'm going to specify the matrix using a text representation I type in at Bank of F to show I'm listing the bank elements of the F matrix clicking ok again produces results we want again we see the restriction we specified at the top followed by the estimated parameters and finally the computed matrices note the big restrictions on F implying s can be imputed but it is not possible to be composing A or B the third demonstration is taken from section fourteen point six point one of the Martin herd and Harris book here we're using quarterly macroeconomic data namely oil prices US output interest rates and the US consumer price index the underlying var is a three lag model using growth rates or d-logs for oil prices output and CPI and levels of the interest rates we'll use a time trend as an exhaustion aspherical to specify the s bar we're going to impose restrictions such that oil shocks only affect oil prices in the short one interest shocks don't affect output in the short one an inflation and interest shocks don't affect output in the long run we can do this by mixing restrictions on both s and F to specify the oil short-run restrictions I specify that the output inflation and interest columns of s are zero to specify the interest shocks don't affect output except the interest output entry to zero finally to specify the long-run restrictions I set the CPI and interest columns of output to zero in the F matrix note that specifying these long-run restrictions in F is equivalent to imposing more complicated linear restrictions on s clicking OK to produce the output reveals similar output to the other models note however that EU has reported both the specified restrictions of s and F as well as the linear restrictions in S implied by the given restrictions of F our final demonstration comes from section fourteen point six point three of the book and again handle some macroeconomic variables but this time from Australia we have quarterly real GDP growth interest rate growth return on Australian equities inflation rate growth and return on US equities well estimate a bar with two legs and a couple of date based dummy variables as exogenously rules you when we estimate the structural factorization we will impose a number of long-run restrictions including the restoration of financial assets are priced in the long run at present values which transfers into the restriction to the 3/2 element of the F restriction matrix is equal to the negative of a 2/2 element will first impose the GMO restrictions by starting with one of the preset options for F then modifying it slightly you to impose the equality without a further check restriction the output again displays our restrictions and the estimated parameters now that the estimated coefficients do indeed follow the quality restriction we imposed [Music]

Channel: IHSEViews

Views: 30,456

Rating: 4.9245281 out of 5

Keywords: EViews 10, EViews, structural VARs, restrictions

Id: _nGkJzDlPY4

Length: 8min 44sec (524 seconds)

Published: Tue Jun 20 2017