EViews 6 Patch Installer
To determine if your installed EViews is out-of-date, you can compare the "build date" of your version of EViews with those posted below. Your version's build date is displayed when you select About EViews... from the EViews Help Menu.
The EViews 6 update executable may be used to update your currently installed EViews 6 to the most recent shipping version. The update installer not run unless you already have EViews 6 installed on your machine.
February 07, 2011
- Fix for invalid registration attemps.
Fix equation string representation in tobit specifications.
Fix for Quandt-Andrews test in non-linear equations. Also increased the number of breakpoints tested by 1. Fix for incorrect Wald statistic being used in the Quandt-Andrews test with White standard errors. Fix for a crash when passing a large group in as the date series in a page re-structure.
Increase in the number of equations supported in Model objects. Fix for a crash in some dynamic panel GMM specifications. Fix for an errant log of a negative number error message in some @cum functions.
Fix for an errant minus sign showing on some zero values in histograms. Fix incorrect degree-of-freedom in calculation of selected by-category median equality tests when specified using multiple factors. Fix for an obscure bug with panel box plots. Fix for recently introduced bug in White heteroskedasticity test performed via dialog. Fix for export sample when making residuals from an equation with ARMA errors and an adjusted sample.
Fix for a bug in saving graphs involving shading and line objects made from linked series. Fix for a bug in storing to existing individual database files. Fix for a bug in storing irregular data to existing database files.
Fix for a slight bug in the calculation of the White Heteroskedasticity test in cases involving dropped regressors. Added the NOLIMIT option to by-group procedures. Fix for a bug in user-specified X12 procedures. Fix issues with writing to SAS portable files. Improve compatibility with kdb+ ODBC driver.
Changed the calculation routine of correlations as part of principle components estimation when pairwise samples are used, to match that used in other routines.Fix for an error introduced by an earlier patch that caused periodic compilation errors of expressions with many spaces. Fix for an error when opening workfiles containing groups created with an @expand command. Added an error message to notify users of a non-terminated for loop.
Fix for a bug when using the name command on multiple series graphs. Fix for a bug in scatter plots with regression lines which caused a horizontal line to be shown rather than the true regression line in rare cases.
Fix for a crash when using certain series functions on a group with more than 10 series inside it. Various panel/pool estimation fixes. Fix for a bug where freezing the spec view of a State Space, System, Model or LogL object would create the wrong object. Fix for a bug in running X12 from an external source file.
Fix for a bug in fetching multiple groups from a database via the command line. Fix for a bug in the Database Storage options dialog. Fix a bug in the calculation of cumulative standard deviation and variance routines.
New warning message notifying the user of the incompatibility of the X11 routine with 64-bit operating systems. Fix for a bug in certain views and procedures performed on differenced series. Fix for a bug in Factor Analysis correcting variance weighting of data when computing factor scores for models estimated from the covariance matrix of the original data.
For Windows XP SP2 and later, EViews 6 now supports registration via proxy servers defined in Internet Explorer. Proxy server settings defined in Internet Explorer (see Microsoft Help & Support link for information on how to do this: http://support.microsoft.com/kb/135982) will now be used by EViews 6 when performing registration.
Improved compatibility with .SAS files written by Data Ferret. Fix for a bug in the Make Model process from a panel equation with fixed effects. Increased the accuracy of the normal distribution function. Fix for a bug in undated panel restructuring whereby the indices could become switched. Fix for a bug in the single letter database type specification via the fetch command. Fixed sizing issues when printing objects from a program. Fix for a bug in Recursive Least Squares performed on weighted equations.
Various graph related bug fixes. Removed the degree-of-freedom correction that was being made in VAR residual normality tests. Fix for a bug that stopped the F-statistic from showing in equations. Fix for a bug in Make Model from equations or systems with ARMA coefficient restrictions. Fix for a bug where ‘wfsave(type=odbc)’ command would open the popup “Specify table” popup with the incorrect default table name (if a table name was specified in command). Fix for a bug where ‘pageunstack’ command would open the ‘Workfile Unstack’ popup with the wrong defaults for new workfile name (if specified in command). Added functionality to the Define System dialog.
Fix for a display bug.Fix for a bug in the File Open dialog where changing the file type would have no effect.
Fix for a bug in graph options involving linked series. Fix for a bug in copying pre-EViews 6 graphs. Updated FAME support code providing the ability to write to newer FAME server databases. Fix for certain issues with the dialogs for breakpoint tests. Fix for bug in univariate ARCH backcasting at non-default settings. Fix for rare errors/crash in freeze command. Fix for a bug in programs with sub-routines that may, in rare circumstance, cause EViews to randomly crash after the program is run. Fix for a bug in GARCH forecasting when using "variance targeting" as an estimation restriction. Slight tweak to GARCH estimation. Fix for certain graph print commands causing EViews to crash. Fix for categorical graph commands. Miscellaneous other fixes.
Fix for global graph labeling position options not being applied. Fix for row-wise standard deviation and variance calculations. Change to allow entry of username and password in dialogs for FAME databases. Fix to the normal distribution random number seeding. Fix for option handling on system re-estimation via command line. Various fixes/changes to the multivariate ARCH interface.
Various changes/fixes to the Make Distribution Data functions. Fix for a crash when copying database links whilst in Workfile Details view. Fix for a bug that would cause re-defined formulas to be un-deleteable. Fix for a crash when choosing Quick->Sample from a program. Fix bug in writing Gauss files from the dialog. Modified panel cointegration error messages and error handling. Update for multivariate ARCH estimation, allowing models with no constant in the mean equation. Miscellaneous bug and error fixes for multivariate ARCH. New command to access fixed effects from a panel equation (@effects). Remove spurious error message when performing series classify by quantile with large numbers of tied values. Improved handling of ambiguity in series classification for user specified limit points with repeated limit point values. Fixed a bug causing inexact reading of integers stored in scientific notation.
Fixed a bug with encrypted programs being shown in EViews 6. Fixed a bug causing inexact reading of integers stored in scientific notation.Increase the maximum number of equations allowed in a Model object to 300,000. Correct various spelling mistakes. Correct mislabeling of the parameters of the Pareto distribution in EDF tests and distribution plots.
Fix for a bug in the @mov or @m moving functions that require two series as inputs where incorrect values were being calculated for observations at the start of the workfile. Fix for a bug when producing Kernel distribution data from multi-series groups. Fix for a bug in exporting lagged series. Improved syntax colouring performance - especially on large programs. Fix for a bug in dynamic forecasting of equations with certain transformed dependent variables. Fix for a bug in co-integration estimation and testing, and unit root testing of series with a seasonal difference. Fix for rare errors when reading certain Datastream series. Fix for a bug when writing date columns containing non-zero second information to SQL Servers over ODBC. Fix for a bug when writing certain columns of integers to MySQL over ODBC. Fixed a bug when creating samples based upon series stored in a database. Disallowed Pool unit root and co-integration tests in Panel workfiles. Fixed a bug causing NAs to be reported in Pearson Pairwise Correlation calculations.
Fixed a bug in fetching certain series from Datastream which could lead to an erroneous "invalid start date" error message..Added command line option for model solves. SOLVEOPT(STRUCT=t) now tells the model solve to ignore ARMA terms and use only the structural part during the solve. Adjustments to data types of numeric columns when saving a workfile via ODBC. Added internationalization support when using the clipboard to copy and paste non-English characters. Applies to both formatted and unformatted text. Fix bug in multivariate CCC GARCH model when exogenous variables with common coefficients but no exogenous variable with individual coefficients are included in the variance equations. For multivariate CCC GARCH, improve algorithm in the generation of the starting coefficient values for exogenous variables. Change to the specification of the White test for heteroskedasticity in equations containing ARMA terms or weights.
Febuary, 2008Fix for stepwise dialog in pre-Windows XP Operating System. Add command line option to save panel cointegration results to a matrix. For (Pool) Pedroni panel cointegration test, fixes 1) user-specified maximum lag is ignored; 2) ADF statististics might be incorrectly calculated when NA is present in one or more of the regressors; and 3) intermeidate test results are displayed in the incorrect order when one or more cross-sections are dropped from the test. Fixes for bugs in Pedroni panel cointegration tests where probabilities are errorenous and ADF are incorrectly calculated. Fixes copying of equations between workfile that are then loaded into models. Fix to remove "too many columns" error message when writing to MySQL databases. Improve behaviour when overwriting tables in Oracle databases. Improve detection of cells containing dates when reading from Excel files.
January, 2008Fixes incorrect placement of daily data in workfiles when fetched from Econwin. Modify behavior when encountering reduced rank covariance matrices in panels and pools to retain the covariance while offering an improved warning. Fix for distribution plot label options not being read when entered by command. Fix for makpcomp options not being read when entered by command.
December, 2007Fix for the Chow Breakpoint Test where the Wald statistic was incorrectly determined to be singular in some non-linear equations. Fix for Stepwise Regression where very small sample sizes generated an erroneous "square root of a negative number" error message. Update the IPS panel unit root test to display probabilities for individual ADF test results. Makes metafile tick spacing match onscreen graph tick spacing.
November, 2007Fix problem in line graphs when connecting over missing observations. Added "DROP" option to the Graph::options command, which drops excluded observations from the graph scale. Fix for calculation of medians on data containing only NAs. Fixed a crash in some FIML calculations. Fix for certain nested moving functions. Fixed string buffer overflow encountered when estimating exceedingly long and complicated panel equation specifications. Fix for a bug in specifying areas of a graph to be shaded by command. Fix for an occasional incorrect error message in the omitted variables test in pool equations when specified by command. Fix for a crash in model solves when the specified scenario has been deleted. Fix problem with "stop solve on missing data" in models. New Feature: Obtain and store the data used in the creation of distribution type graphs, including kernel densities. Fix to work around occasional incorrect start dates reported by Datastream when fetching some series. Replace test for reduced rank covariances in panels and pools with scale invariant version. Fix crash in RESET test when testing uninitialized equations. Fix for graphs created by the OUTPUT(F) command. Fix for errors when saving workfiles to ODBC databases where series contain only small values.
October, 2007Fix for bug with ADDINIT command in models after a solve which may be applied over the wrong sample, sometimes leading to an "unable to find appropriate interval" error message. Fix problems with combining separately generated historgram graphs into a multiple graph. Fix in systems the DW statistics incorrectly labeled as the probability of F-stat. Fix for problems with FIRSTOBS and LASTOBS options in ascii file reading. Support for reading STATA 10 files. Removal of unnecessary prompts for user name and password when reading from FAME server databases. Fix for crashes when writing workfiles in Givewin, RATS, or TSD format. Fixed erroneous repporting in MA forcast when out-of-sample expression errors, such log of a negative number, occur. Fix to prevent loss of last period of data when fetching weekly frequency data into a workfile based on a different day of the week. Fixes crash in serial correlation LM test in equations with MA errors. Fixed an issue with merged graphs not showing lines correctly. Prevent a crash caused by insufficient observations during certain robust standard error procedures in pools/panels. Fix for table print problems. Fix for reading of foreign files where multiple column names begin with the same 24 characters. Correct incorrect returns of some VAR member functions for restricted EC estimations. Overiding the default date intervals on graphs. Fix redirect output to a file path longer 32 characters. Fix characters. Fix memory leak on pagecontract on a panel.
Adds keyboard shortcuts CTRL-O, CTRL-S CTRL-N, CTRL-P, CTRL-A and F2. Changes the default printing option for tables to "print entire table" instead of "print selected cells"; this option can be overwritten in the print dialog. Fixed an issue with automatic lag length selection in unit root tests. Fixed an issue with graph axis borders and graph axis label formats not showing properly when used together. Fixed a bug with database fetches using the (d) option when no default database exists and the default database is included in the search order. Prevented an erroneous error message during panel unit root tests when some cross-sections have less than 4 observations.
Fix for a bug that would cause EViews to hang indefinitely when graphing certain series created by expression. Fixed a problem with the wildcard operator when page stacking pages with more than 32,000 series. Fix for incorrect handling of the graphing of linked series which were linked from an unstructured workfile page, which could result in a blank graph being drawn. Fix for a bug in Newton and Broyden solvers causing erroneous error message if the solve involved a series called INF. Fix for a bug in the calculation of the weighting matrix for GMM with a HAC covariance matrix. Fix for a bug with database fetches and some frequency conversions introduced in the August 21st 2007 patch. Fix for a bug that can occur when freezing a model's message view. Changed behavior of frequency conversion between two two-dimensional panels so that frequency conversion occurs on a per-cross-section basis. Fixed very rare hang during model solves for certain models. Improved panel unit root tests - up to five times faster with auto-lag determination and up to twenty times faster for fixed lag length specifications. Also, the algorithm for auto-lag determination when the data contains missing observations in each cross-section was improved, resulting in fewer dropped cross-sections. Fix display bug in sample for Quandt-Andrews unknown breakpoint test. Fix lag handling for group descriptive statistics with individual samples in panel workfiles. Improve starting value determination in computation of quantiles of the gamma distribution for select parameter regions. The difference operator with a seasonal difference term and high order ordinary difference terms did not calculate properly.
Fixes matrix @csum function and the series @mcov, @movs and @mcovp functions. Changed the precedence of unary negative sign and raised to the power (^) in scalar operations to match that of series operations so that for example -3^2 = -9 instead of 9. Fixed a bug in rare cases involving combinatorial stepwise with two regressors being chosen, where one of the chosen regressors was last in the list of regressors to be chosen from. Fix problems with Makegroup, Maketable and Makegraph procedures of models when "Actuals" is the comparison scenario. Suppress invalid LR test statistics in panel estimation with EGLS weights iterated to convergence. Existing F-statistic values are unaffected. Minor cosmetic change in Goodness of Fit output for Factor Analysis. Fixes problem fetching Econwin data which include observations dated prior to 1900. Fix problem in ARCH estimation when AR or MA term is included in the mean equation and analytic derivative is used. CLOSE command now works with workfile or program that contains spaces in its name. Fix problems with relative paths in INCLUDE statement in programs. Fix problems with applying White heteroskedasticity test on weighted least square estimation. Fix problems with relative paths in include statement in programs. Fix fileloads which include autoseries with @MOV functions that take expressions as arguments.
Fix crashes when defining a new formula based on non-existing series. Fix a problem with taking the determinant of large matrices which may cause crashes. Dramatically increases the speed of fetching data from a Haver Analytics database when the database window is open on the screen (See DBOPEN). If database window is not open fetch speed will be slower. Fix crashes in HFETCH for series exceeding 2000 observations. Change to symbolic derivative code to avoid applying lags to arguments of descriptive statistics functions. Fix add init procedure in model containing autoregressive errors. Fix for model solution when all feedback variables are excluded from a simultaneous block. Extended length of date specification in Chow test. Fix a problem of incorrect object type printed from a workfile view. Modify descending order sort so NA's are placed at the bottom of the sorted series in order to maintain compatibility with version 5. Modify define-system proc to place parentheses around regressor terms in cases where they are needed. Fix to detect missing exponent so that strings such as "3d" or "5e" are not treated as valid numbers. Update to PC Give and TSP portable formats to support 64 character names. Fix some display and dialog problems for sample with very large if conditions. Fix a problem in "space between bars" option in mixed line-bar graphs. Additional options added to model solve command. Fix a crash when redirecting text to a text file. Fix a problem while fetching some daily data from Datastream. Fix a bug in histogram graphs with user-defined bin widths in rare circumstances. Fix a crash in printing from the command line or programs under unusual conditions.
Fix a crash where an error takes place within a program loop when "allow multiple errors" is set to greater than one. Fixes to model solution to improve handling of cases where an equation residual is invariant to its own endogenous variable. Adjustment to numeric derivatives in model solution to reduce problems when an endogenous variable is converging to zero. Improve file IO and bug fix on sparse table objects. Fix a bug in "make residuals proc" in panel equations for two-way fixed effect models with unbalanced data; the estimation results are not affected by this change. Fix problems with R^2 calculations for some weighted panel and pool equations. Correct errors in identification of zero residual observations in subsampled quantile regression estimation. Fix a problem with samples being set incorrectly but reported correctly with irregular integer dated data. Fix for @CLOGNORM function. Fix incorrect handling of NA's across workfile seams in some expressions containing lags or leads. Fix an incorrectly error message regarding sample range given, under some conditions, in a @-statistics command, such as @mean. Fix for a rare crash when solving models. Fix problem with database queries involving Factor objects. Fix a problem with graph option settings in programs. Fix for spurious "ODBC driver not found" error messages. Improve labeling in Links and Formula Series after links have been broken. Fix error messaging for fixed effect tests in panel models with no fixed effect. Fix a problem with workfiles saved in single precision if original data contains numbers greater than allowable single precision value. Fix for error not being set when maximum iterations reached when solving models. Fix for problem fetching formulas from database where auto search is on in the database registry. Update X11 modules with XP compatible ones.
Fix periodic crashes while using mouse wheel in Terminal Server session. Fix crash in "full sample asymmetric" band width filter. Fix periodic problems in line graph on data containing link that has option set to native frequency including axis extension or categorical graphs. Enable scrolling in VAR impulse response dialog. Fix for Windows N.T. 4.0 incompatibility. Handle lagged series in quantile functions. Enable subroutine in programs to recognize factor or spool objects. Add option to handle the treatment of auto-series-generated errors in views and procedures.
Fix problems when resizing workfiles when there are open database link windows. Properly process "No" (do not continue) selection in dialog warning users about estimating instrumental variables specifications with excessively large number of instruments. Fix issues with ARCH-M models when numeric derivative option is not selected. Fix crash when attempting to create an object whose name was used by a previously deleted series which was also part of a group. Fix inability to load workfiles containing group auto series based on "@r" group functions. Fix assorted display issues. Fix command window sizing issue, where, on some computer systems, the window size increases slightly on each successive opening of EViews. Fix @LOWER command to return correct capitalization. Correct inability to cancel EViews' shutdown dialog. Fix Breusch-Godfrey serial correlation LM test for TSLS. Fix in reading fixed precision numeric data from ODBC database. Fix for text truncation in dialogs. Fix @coef to return PDL coefficients in PDL estimation. Fix auto fetch from database.