EViews Related News
EViews Version 4
30 March, 2002
EViews Version 4
Whether you use EViews for general statistical analysis, time series estimation and forecasting, large scale model simulation, presentation graphics, or simple data management, Version 4 provides new features that expand your capabilities.
Estimation and Estimation Object
EViews 4 provides a choice of analytic or numeric derivatives for most nonlinear estimation techniques. Analytic derivatives are automatically calculated by EViews new calculus engine. You may examine the analytic expressions for the derivatives of your specification and save gradients of the objective function for further analysis.
In addition, changes have been made in convergence testing to improve estimator robustness. More informative output keeps track of your estimation settings. Expanded global options allow you to define default settings for estimation procedures.
EViews now handles coefficient restrictions in vector error correction (VEC) models. You may impose and test linear restrictions on the cointegrating relations and/or adjustment coefficients. This allows you to choose your own normalization for the cointegrating vectors and to impose weak exogeneity restrictions. EViews displays LR test statistics for binding restrictions.
There are a number of new views in VARs that allow you to examine the structure of your specification. With a few clicks of the mouse, you can display the inverse roots of the characteristic AR polynomial, perform Granger causality and joint lag exclusion tests, evaluate various lag length criteria, and view correlograms and autocorrelations.
VARs also perform various multivariate residual based diagnostics. In addition to displaying pairwise cross correlograms, EViews computes serial correlation (portmanteau and LM), heteroskedasticity (White’s), and normality (with various options for orthogonalization) tests.
EViews VARs now provide various options for the impulse response specification. In addition to the default Cholesky factorization, you can now set the impulse to one-unit or one- standard deviation residuals (ignoring correlations), generalized impulses (Pesaran and Shin 1998), structural factorization, or a user specified vector/matrix. Monte Carlo standard errors are now available for variance decompositions.
Structural factorizations (VARs) may be estimated by imposing short-run (Sims 1986) or long-run (Blanchard and Quah 1989) restrictions. Restrictions may be specified as in text form, or as patterned matrices. Over-identifying restrictions may be tested using the LR statistic reported by EViews.
The state space engine has been completely rewritten to provide you with greater power and flexibility. A brand new syntax allows you to specify much more general state space models than was possible in EViews 3.1. Among other things, we now allow exogenous variables in the state equation and fully parameterized variance specifications.
New procs and views give you better access to powerful filtering and smoothing tools so that you can view or generate one-step ahead, filtered, smoothed signals, states, errors, etc.
Sophisticated forecasting procedures provide easy-to-use tools for in- and out-of-sample forecasting using n-step ahead or smoothed values.
In addition to the default exponential smoothing backcasting methods for initializing ARCH variances and residuals, EViews 4 now allows you to initialize these values to the unconditional variance. New EViews programming examples show how you can use the EViews Logl object to estimate multivariate GARCH models of up to three-variables.
Model Solution and Analysis
The model object, which is used for simultaneous equation simulation and solution, has been completely revised. A powerful new graphical interface allows you to view the structure of your model organized by variable or by equation. Variable dependencies and the block structure of the model’s equations are displayed with a simple mouse click.
Simulation/solution options have been to provide full stochastic simulation. You may specify alternative assumptions about residual uncertainty, coefficient uncertainty, and exogenous variable uncertainty. It is now possible to generate confidence intervals for your multiple equation forecasts. Multi-period policy multipliers, including confidence intervals, can easily be generated and displayed graphically.
A new Newton solver has been added to complement the existing Gauss-Seidel solver, allowing solution of models that could not be solved in previous versions of EViews.
Forward solution (currently unavailable with stochastic solution) allows you to solve for model consistent expectations.
In addition to expanded solution options, a variety of other features have been added to models. Multiple solution scenarios are easily managed and their output compared graphically or in tabular form. Greatly expanded add factor support, including equation normalization, is now provided. Simple control problems may be solved where the values for an exogenous control variable are found so that an endogenous variable achieves a user specified target.
Enhanced Graphics
The EViews graphics engine has been completely revamped to provide the user with greater control over the display of graphics. Users can now control aspect ratios and provide independent choices of line type, line thickness, symbol, and color. New options allow the user to customize axes scaling, tickmarks, labeling and label placement. Every option can be set either interactively through the new graph option interface or in programs using an expanded set of graphics commands.
New graph types allow you to plot multiple pairs of series in a single XY-line graph or scatterplot. New types have also been added for high-low-(open-close) graphs, spike graphs, and points with error bars. A specialized graph allows quarterly and monthly series to be displayed by season to assist in the visualization of seasonal patterns.
Transferring graphs to other programs has been improved with support for Windows Enhanced Metafiles.
Statistics
- Principal Components
New principal components tools can be used to summarize your data or to analyze colinearity. The principal components view of a group of series finds the linear combinations of the series with the largest or smallest possible variance. You may also ask for the principal components of a sample covariance or correlation matrix. The eigenvalues, eigenvectors, and any number of principal components can be easily stored as named series in the workfile. - Distribution Tests
EViews provides built-in Kolomogorov-Smirnov, Lilliefors, Cramer-von Mises, Anderson-Darling, and Watson empirical distribution tests. You can test whether your series is normally distributed, or whether it comes from, among others, an exponential, extreme value, logistic, chi-square, Weibull, or gamma distribution. You may provide parameters for the distribution, or EViews will estimate the parameters for you. - Independence Tests
A new series view carries out the BDS test for independence. The BDS test is a portmanteau test for time based dependence in a series. It can be used for testing against a variety of possible deviations from independence including linear dependence, non-linear dependence, or chaos. - Resampling Tools
EViews new resampling tools may be used as the basis for Monte Carlo techniques such as nonparametric bootstrapping and permutation tests. The resampling tools allow you to draw a random sample with or without replacement from observations in a series or group of series, or from the rows of a matrix. The series/group resample provides various options for missing value handling. - Random Number Generators
Two new random number generators have been added: the L’Ecuyer (1998) "optimally" combined multiple recursive generator and the Matsumoto and Nishimura (1998) Mersenne Twister. - Distribution and Kernel Density Plots
You can now save the results from your distribution and kernel density plots in matrix objects.
Seasonal Adjustment
- Census X12-ARIMA
EViews 4 includes support for the U.S. Census Bureau X12-ARIMA program. As with the existing X11, EViews provides a front-end interface to the official Census release. The new interface provides you with easy-to-use, menu-driven access to the most commonly used X12 features, and a new command interface to the full power of the Census X12-ARIMA program. - Tramo/Seats
Tramo (Time Series Regression with ARIMA Noise, Missing Observations, and Outliers) performs estimation, forecasting, and interpolation of regression models with missing observations and ARIMA errors, in the presence of possibly several types of outliers. Seats (Signal Extraction in ARIMA Time Series) performs an ARIMA-based decomposition of an observed time series into unobserved components. Used together, Tramo and Seats provide a popular alternative to the Census X12 program for seasonally adjusting a series. The two programs were developed by Victor Gomez and Agustin Maravall.
EViews provides a convenient front-end to the Tramo/Seats programs as a series proc, using both a graphical and command line interface.
Data Management
- Displaying and Editing Data
A new spreadsheet feature allows you to display and edit your data in difference (1 period or 1 year) or percent difference (1 period, 1 year, or 1 period annual rate) forms. Simply open a series or group spreadsheet, and tell EViews that you want to display your data in "1-period percent changes" and EViews will automatically change the spreadsheet display to show the differences in percent terms. You may then edit the percent change for an observation and all of the underlying series data will be modified to reflect this change.
Dated Data Tables have been enhanced to allow user placed heading rows. - Workfile Extraction
You can now create subsets of an existing workfile. Simply specify a sample, the types of objects you wish to keep, and a description of the names of objects to be kept and to be dropped. EViews will create a new workfile containing the objects and data that you selected. - Database Support for RATS, TSP, GiveWin, and Aremos TSD Files
EViews now supports RATS, TSP, PcGive, GiveWin, and Aremos TSD files through the same interface provided for EViews databases. For example you can open a RATS file and copy-and-paste series into an EViews workfile or database. All EViews database operations, reading, writing, querying, etc., can be applied to these file formats. - Enterprise Edition Support for FAME, DRIBase, and Haver Analytics Databases
As part of the EViews Enterprise Edition (an extra cost option over EViews Standard Edition) support is provided for proprietary data formats of commercial data and database vendors. You can access FAME local and server based databases, Standard and Poors DRIBase databases, and native Haver Analytics DBX databases. The same, easy to use, EViews database interface has been extended to these data formats.
Programming
EViews programming language has been enhanced with better error handling, new object data members, and a wide range of new functions: new statistical distribution functions, functions for computing elapsed time, functions for determining current date and time, functions to retrieve installation and working directory paths, and functions that return the number of observations in the active workfile range and sample.
Online Documentation
The EViews Help System has been expanded with more extensive discussion of estimators, screen shots, and additional hypertext links. In addition, the entire EViews User’s Guide and EViews Command and Programming Reference are provided in Adobe PDF format (along with Adobe Acrobat Reader). Both manuals are extensively hypertext linked, making it easy to find the information you need.
Order EViews 4 from Hearne online.