meactohn2003

可能你应该先用ｌｉｂ命令建立一个ｌｉｂｒａｒｙ文件吧，因为别人把用到的程序做成library文件。

http://www.texlips.net/svar/



Author: Anders Warner



Structural VAR is an extensively used method in Macroeconometrics, especially in empirical monetary economics. which is ntroduced by sims(1980) to macroeconomics



For example, on identification strategy of SVAR, eviews can not simultaneously impose long- and short-run restrictions; results on stability test are also reported in the estimation output.



More information on what it can do please visit the above website. Read the help file to get started using it.



One remark for windows XP, you need to right click the desktop, then click property and switch to windows classic.



I am learning on use it, and anyone who would like to learn or discuss on it may send an email to me.



xukuang910@hotmail.com



Features:

Estimate VAR models using Maximum Likelihood or analyse them based on Bayesian methods.



Perform cointegration analysis.



Estimate alpha, beta and C under general linear restrictions.



Estimate structural VARs with identifying restrictions on the contemporaneous and/or long-run effects of shocks.



Allows for over-identifying restrictions on the structural model including over-

identifying long-run restrictions.



Can estimate all parameters of the system simultaneously.



Can bootstrap over-identification tests and parameters for structural VARs including impulse response functions.



Can forecast levels and growth series of the endogenous variables and functions of linear combinations of these.



Highly configurable with a large selection of options.



Uses model files which allows fast switching between various basic specifications.



Compute a wide range of specification tests if asked to do so!



Import data from text files, MatLab MAT files, Lotus 1-2-3 spreadsheets, and Excel spreadsheets (under either MatLab 6.x or later, or for stand-alone executables when they have been compiled using MatLab's compiler for MatLab 6.x or later).



Export data to text files from lines and patches (confidence bands and histograms) in the stand-alone release.



Allows for partial systems.



Compute Bartlett corrected trace tests.



Can compute p-values for trace tests.



Distributions for trace test in partial systems are included for a maximum of 12 endogenous variables and 11 exogenous I(1) variables.



For partial systems with only exogenous I(0) regressors, p-values and critical values for the trace test are provided in models no additional deterministics. That is, models with restricted and/or unrestricted constant and/or linear trends as well as with or without centered seasonal dummies are covered. The approach has been suggested by Boswijk and Doornik (2005; see help file) and relies on using the gamma distribution with the appropriate mean and variance.



The distributions for the trace test with or without I(1) exogenous variables can be simulated from within Structural VAR.

Can bootstrap LR trace and Bartlett corrected LR trace tests for the cointegration rank using either "parametric" (draw normalized residuals from a Gaussian distribution) or "non-parametric" (draw new residuals from the estimated residuals using a uniform distribution over the estimated residuals) construction of bootstrapped series.



Can bootstrap "parametrically" and "non-parametrically" the free parameters of the cointegration vectors once these have been identified as well as the LR based test for the restrictions on the cointegration space.



Structural VAR can now bootstrap all parameters on the short-run dynamics, i.e., alpha (on cointegration relations), delta (deterministic variables), Gamma (lagged endogenous variables in first differences), Psi (exogenous I(1) variables in first differences), and Phi (exogenous I(0) variables). If the model contains restrictions on alpha, the test statistic for these restrictions can also be bootstrapped from the same dialog.



Compute lag order, weak exogeneity (wrt alpha, beta), Granger non-causality and common cycle tests.



Can bootstrap specification tests, weak exogeneity, Granger causality, lag order tests.



Compute various formal parameter constancy (e.g. Nyblom and fluctuation) tests and display them graphically.



The distributions of the Nyblom mean and supremum tests can be simulated from within Structural VAR.



The asymptotic distributions of the Nyblom mean and supremum tests are included for the models with a restricted constant, with an unrestricted constant, with a restricted linear trend, or with an unrestricted linear trend.



Can bootstrap the Nyblom test, the fluctuation tests for the non-zero eigenvalues and the Ploberger-Krämer-Kontrus fluctuation tests for the non-cointegration parameters both "parametrically" and "non-parametrically".

Bootstrap exercises for the Nyblom tests allow for power studies (using a fixed alternative) as well as Monte Carlo bootstraps.

Residual analysis with graphics.



Display cointegration relations and permanent and transitory components based on Beveridge-Nelson decomposition.



Compute generalized impulse responses for the levels and the cointegration relations along with asymptotic confidence bands and display them graphically.



Display historical forecast error decompositions.



Variance decompositions for structural models.

Graphs of the estimated free coefficients in the identified cointegration space. These graphs are given in 2-D and 3-D format, with 1 or 2 beta coefficients graphed in a region around the maximum against the value of the log-likelihood function.



Quick view of estimated parameters and various test results.



Save estimated time series into files.



Model selection help function: use information criteria for combinations of lag order and rank (if cointegration analysis is selected).



A rich set of graphics editing tools including data editing.

Save graphics into a wide range of file formats; the stand-alone release relies heavily on ghostscript and pstoedit.



The output file can either be produced as formatted plain text or formatted LaTeX. In the latter case, Structural VAR can display previews of the output in dvi (device independent), ps (PostScript), and pdf (Portable Document Format) provided that the needed programs for compiling into and interpreting these formats exists on your hard drive (or network) and Structural VAR has been informed about their location.

http://faculty.washington.edu/ezivot/econ584/time_series_econometricians.htm







Time Series Econometricians





Jushan Bai, Boston University



Anil Bera, University of Illinois at Urbana-Champaign



Peter Boswijk, University of Amsterdam



Mark Dwyer, UCLA



Robert Engle, UC San Diego



Jesus Gonzalo, Carlos III



Philip Hans Franses, Erasmus University Rotterdam



James Hamilton, UC San Diego



Bruce Hansen, University of Wisconsin



Andrew Harvey, Cambridge University



David Hendry, Oxford University



Oscar Jorda, UC Davis



Chang-Jin Kim, Korea University



Jan Kiviet, University of Amsterdam



Frank Kleibergen, University of Amsterdam



Gary Koop, University of Edinburgh



Andre Lucas, Free University of Amsterdam



Roger Moon, UC Santa Barbara



Serena Ng, Boston College



Maurice Ooms, Erasmus University Rotterdam



Peter C.B. Phillips, Yale University



Simon Potter, Federal Reserve Bank of New York



Norman Swanson, Texas A&M University



Dick van Dijk, Erasmus University Rotterdam



Herman van Dijk, Erasmus University Rotterdam



Mark Watson, Princeton University



Halbert White, UC San Diego



Time Series Courses

UCLA



UC Davis



UC Santa Barbara



University of Washington

看到一个很不错的关于Ｓｔｒｕｃｔｕｒａｌ　ＶＡＲ的介绍,和有兴趣的朋友分享，很简洁紧凑。

ｓｔｒｕｃｔｕｒａｌ　ＶＡＲ在宏观计量中学到，尤其是经验货币经济学的主要方法之一。



Eric Zivot

Notes on structural VAR modeling

google搜一下（１８页）

http://www.eco.uc3m.es/jgonzalo/teaching/timeseriesMA/zivotvarnotes-reading.pdf

做出来的Ｓｌｉｄｅｓ：

http://faculty.washington.edu/ezivot/econ584/notes/svarslides.pdf

http://faculty.washington.edu/ezivot/econ584/notes/svarslides2.pdf

这里边还蛮有学问的，长了不少见识，红外线监测器看到的比较多

人气旺了不少，可喜可贺

是啊，很久没看，看了一下短消息，里边有一条就跑过

来了，

不好意思没看到被移动到这里来了，不过现在不需要了多谢两位！！

Hamilton 是在Ｐｒｉｎｃｅｔｏｎ吗？

Are we talking about Truncated regression model? I think there is no identifiabilty problem here, we are not estimating model coefficients from data!! r is known .And here of course the functional form is linear, otherwise   z can not be truncated norm distribution(because x,y are normally distributed). Perhaps the question should be specified more clearly.

Truncated distribution is a type of conditional distribution: density f(x|x>a)=f(x)/P(x>a) truncated from below,or from above f(x|x<a)=f(x)/p(x<a), for example, f(x) can be any distribution, uniform distribution, Normal distribution, Possion distribution.





In our problem, the density function of z has a density form as following:

f(x|b<x<c)=f(x)/p(b<x<c)=f(x)/G(c)-G(b)

here f(x),G(x) denote the density function, and distribution function of normal distribution(or we take it as standard norm distribution).





z correlates with w only because both depend on x, assume z and w are linear function of x,y:



t=a(1-r)x+dy

w=erx

where a,d,e are coefficients.We obtain z by truncating t from both below and above(!!!!),

that is,

z=t|b<t<c

Without loss of generality, or for simplicity, we assume x,y are distributied as N(0,1)

The t distributed as N(0, (a(1-r))(square)+d(square))

Firstly, we compute the the covariance,



Cov(z,w)=Cov(t|b<t<c,erx)=Cov(a(1-r)x+dy|b<t<c, erx)



To compute the correlation coefficients we still need the standard deviation for z and w,

Var(w)=(er)(square)Var(x)

Var(z)=var(t|b<t<c)=E(z(square))-(E(z))(square):

1) E(z(square))=E(t(square)|b<t<c)=(integral)z(square)f(t)/p(b<t<c)dt     b<t<c

2)E(z)=E(t|b<t<c)=(integral)z(square)f(t)/p(b<t<c)dt         b<t<c



You may compute these by yourself, a little complicated but Probability I. Hopefully it's right.

Note that the truncated mean and variance differ from the original norm distribution.The truncated variance is reduced and mean depends on b and c. See for example, Greene, Econometric analysis, Ch22, 757-761, which is available on the website of COS. cos.name ->置顶文章

I studied statistics at Ren min uni of China. We used some English textbooks (or translated) in some lectures, but now I can not remember what are they. Could someone help me to list the textbook(the author, title of the book, publishing year and company ) for the following lectures:   Loss model (taught by Wang Yan), Interest Theory(taught Wang yan), Survial analysis(?), Acturial mathematics(taught Huang xiangyang), Financial management (taught zhou liji) I need it for applicating some Ph.d. programs in Economics next fall. Some Econ department may ask you to make a list of math, statisical, economics lectures that you have done, including the references and name of lecturer. However, It's not urgent for me, someone may do me a favor whenever you have time. Thanks a lot!! Maybe I should put it eslewhere,however, I think people come to this block most frequently.

第一楼说n是样本量但上一楼的n说似乎不是样本量，不知道我的理解对吗——

Gn(x)之间是什么联系？如果没有联系，那n趋于无穷是什么意思？你自己想出来的问题，很厉害，赫赫！

我所学的中心极限定理，

1.Lindeberg-Levy Th.

2.Lindeberg-Feller Th.

3. Lyapunov Th.

1.是最常用的了，条件最强，i.i.d.假设

2., 3.放弃同分布假设，都是基于Trangular array，其中2.　3.的前提分别叫做Lindeberg条件和Lyapunov条件。

证明：1的证明用到特征函数Taylor expansion和continuity Th.　1其实是2的一个特例

2的证明用到Taylor expansion,和依分布收敛的一个等价定理.....

有2了以后，3的证明很简单因为一步可以看出Lyapunov条件意味着Lindeberg条件。

关于Lyapunov Th.的例子有Coupon-Collector's Problem...

Ljapunov

Reference:

Billingsley:Probability and measure 2nd edition 366-379

3nd edition 357-367

问的是经验分布函数吧?

Glivenko-Cantelli Lemma

Let X1,x2,.... be iid as x with (unknown) distribution function F . Let ω be the outcome and be the empirical distribution function based on observations X1(ω),x2(ω).... . Then, as n tends to

infinity tends to 0 a.s.

         

Fix ω Fn converges to F uniformly.