Strictly speaking, operators in EViews are not functions as defined above, though they share a number of characteristics. As the name suggests, operators “operate” on data elements and return results.

There are a few key points to bear in mind:

Many EViews functions may be termed element functions since they takes arguments that correspond to numeric or alphanumeric scalar values, and return a single numeric or alphanumeric value.

EViews generalizes this basic behavior to arguments involving the elements of various data objects including series and alpha series, and the various matrix objects including vectors, matrices, and svectors.

If the argument is a literal or scalar value, EViews will perform the computation and return a scalar value. To take a simple example, consider the log and @log functions, which both compute the natural logarithm of the numeric argument, and the @length and @left functions, which compute the length and return the left-most characters, respectively, of an alphanumeric (string) argument.

The commands,

all compute natural logarithms of the scalar numeric argument. The first command displays the results in the EViews status line, the second creates a scalar object S1, and the third assigns the value to a program replacement variable !S2.

Similarly, for the individual string arguments,

the first command displays the number 4 in the status line, and the third and fourth lines both return the string “When”, assigning the result to a string object, and a replacement variable, respectively.

If the argument is a series, alpha, or group, EViews evaluates an element function using the value for each observation in the current workfile sample, and returns a series or alpha depending on the function return type. In essence, the underlying function is evaluated repeatedly, in a loop over all observations in the sample.

The commands

creates a quarterly workfile from 2001q1 to 2025q4, creates a random series Y, sets the sample to 2010q1 to 2020q4, and assigns the results from the @exp function to the new series TR over that subsample. Note that since the TR series is newly created, it will have NA values for all observations outside of 2010q1 to 2020q4.

Similarly, the commands

initialize the alpha series A to contain the string value “<missing>” for all observations in the workfile, then sets the 2010q1 to 2020q4 subsample of A to contain the results from the @chr function evaluated at the @trend values.

For matrix objects (broadly defined to include svectors) EViews evaluates the function for each element of the object and returns an object of the appropriate type and corresponding size.

Consider the commands

which creates a 3-element vector V1 containing the values 1, 2, and 3, then employs the element function @log to create a 3-element vector LOGV1, with values 0, log(2), and log(3).

Further, the commands

Similarly, the commands

sets up the svector by creating a 3-element vector containing the strings “apple”, “pear”, and “orange” and then uses the element function @length to create a 3-element vector LENS1 with values 5, 4, and 6.

Below we attempt to categorize the element functions into useful groupings. Naturally, the categorizations are not the only ones possible, but we have tried to provide groupings that will aid you in locating a function of interest. Because there are multiple ways to group the information, functions may appear in more than one category.

The following functions provide access to the density or probability functions, cumulative distribution, quantile functions, and random number generators for a number of standard statistical distributions.

Most of the distributions support four different function types:

though some distributions support only a subset of the types. The listing below may also show related functions.

Many of the cumulative distribution functions take an optional argument at the end of the argument list. Non-zero values tell EViews to compute the upper tail of the CDF.

The following functions perform calculations commonly employed in financial analysis.

EViews provides a number of functions for evaluating special mathematical values such as Euler’s constant. For further details on special functions, see the extensive discussions in Temme (1996), Abramowitz and Stegun (1964), and Press, et al. (1992).

References

EViews offers basic statistics functions which compute statistical summaries using the data in various EViews objects. Different functions support different types of source objects, with slight differences in how the function is computed depending on type of source.

Some statistics functions work with objects containing numeric data. Most of these functions will work with any object that contains numbers. You may, for example, compute the mean of series data using @mean, the standard deviation of data in a vector using @stdev, the trend coefficient from the regression of all of the data in a matrix on an intercept and trend using @trendcoef, or the correlation matrix of data in a group of series using @cor.

Other functions work with objects containing both numeric and string data. For example, you may use @max to identify the last sorted value, or @first to find the first non-missing or empty element of series, vector, coef, rowvector, alpha, or svector data.

Crucially, for basic statistics computed using series, alpha, or group data, most EViews functions will default to using data in the current workfile sample. You generally may provide an alternate sample specification, enclosed in double quotes, as the last argument in the function. For vector, svector, and matrix computation using functions, EViews will use all of the relevant data in the object.

Where summaries of matrices are allowed, EViews uses all of the elements of the matrix, by vectorizing the matrix before performing the computation. Special functions are available if you wish to perform individual computations for each of the columns of a matrix ( “Summary”).

For basic descriptive statistics, missing values, either in the form of numeric NAs or empty strings, are ignored.

The following functions facilitate working with all of the observations of series and alpha data in a workfile.

EViews workfile functions provide information about each observation of a workfile based on the structure of the workfile.

These functions may be viewed in two ways:

Since workfile functions are based on the structure of a workfile, they may only be used in operations where a workfile is involved. Workfile functions may be used in statements that generate series in a workfile, in statements that set the workfile sample, and in expressions used in estimating an equation using data in a workfile.

You may compute rolling (moving) statistics that compute summary statistics for a moving window of values in a series object, producing one value for each observation in the workfile. EViews offers functions for computing moving sums, averages, sums-of-squares, variances, standard deviations, minimums, maximums, medians, skewness, and kurtosis for single series, and correlations, covariances, and inner products between series.

The rolling statistics may be divided into two classes. The first set of rolling statistics, with names beginning with “@mov”, propagate missing values. The second set with names beginning with “@m” do not propagate missing values. This class of functions acts like basic summary statistics by removing NAs as necessary.

In general, rolling statistics in EViews are computed on a trailing basis so that a 5-period moving average for a given period is computed using the current and four lagged values of the data.

Note that in contrast to the functions described in “Basic Statistics”, computation of the rolling statistic values does not depend on the workfile sample so that sample gaps will not change the underlying computations and values. The workfile sample only comes into play when you assign the values of the function to a target series.

EViews provides functions for cumulative statistics computed using series data from the beginning or end of a sample to the current observation.

The family of standard cumulatives includes familiar statistics such as means, medians, standard deviations, and variances, to minimums and maximums, quantiles, numbers of observations and missing values.

Importantly, note that these forward and backward cumulative functions default to use all of the data in the workfile. You may provide an alternate sample specification as a double quoted last argument to the function. Observations that are not in the specified sample will not contribute to the cumulative statistics.

In addition to the standard cumulative functions, EViews provides functions for specialized computation of cumulative sums involving first differences of the series. There are three types of functions:

EViews offers a set of by-group statistics which compute statistics for sub-groups of data in the workfile sample, and assign the relevant value to each observation. These by-group statistics may be used as part of any series expression to transform the raw series data into a summary series

The functions take a series expression arg1 for which we wish to compute group statistics and one or more series expressions arg2, defining the values over which we wish to compute the statistics. In the leading case, you will provide a single series expression containing a series or alpha whose values will define the category. More generally, you may provide several series expressions separated by commas (arg2, arg3, ...) whose joint values define the categories.

By default, EViews will use the workfile sample when computing the descriptive statistics. You may provide an optional sample expression or named sample to adjust the computation.

There are two related functions of note,

returns an integer indexing the group ID for each observation of the series or alpha expressions given by the arguments, and

returns a scalar indicating the number of groups (distinct values) for the series or alpha expressions defined by the arguments.

EViews offers a set of statistics which compute summaries using data from all of the series in each row of a group object, and assign that value to the corresponding observation. These by-row statistics may be used as part of any series expression to extract a row summary series from the data in a group.

Note that you may obtain the number of elements in the group rows using the @columns function or using the data member groupname.@count.

Transform By-Row