E-views is the statistical package widely used by economists, researchers and analysts for statistical analysis, forecasting and model simulations (Aljandali and Tatahi, 2018). It is used as a menu-driven object-oriented user interface. It allows writing simple programs in E-views programming. This package offers multi-window design features that allow automation and presentation features. This article explains the comparative analysis of E-views advantages and disadvantages in the time series and panel data analysis. The advantages and disadvantages of the E-views are considered by comparing some of the commonly used statistical packages like STATA and SPSS.
Importing datasheets based E-views
SPSS requires to specify data labels and the numeric values needs to be assigned to each of the string variables. Whereas, E-views allows to import the entire datasheet at a single click and doesn’t require any specification of the labels.
Importing data file into STATA and SPSS accepts a limited set of formats like *.xlsx and *.xlsm. Data files from STATA can’t be imported in SPSS and vice versa. However, E views can import data from multiple databases including SAS, SPSS and STATA specified in * .sas, * sav and *dta formats. In addition to this, it can also import text, binary, HTML file, Gauss dataset file and more.
Output based advantages
The data files in E-views can save the results in a single work file by storing the database in a group format or as individual databases. In SPSS and STATA the output files and the data files are saved separately.
E-views is useful for:
- least squares,
- generalized linear models,
- quantile regression,
- autoregressive conditional heteroscedasticity,
- binary choice and,
- co-integration regression by specifying only the regression equation for the sample period.
In contrast to this, follow the series of different steps and commands for every analysis.
Feature based advantages
E-views offer features like an interpolation of data and frequency filter options through which the missing data can be generated. Due to this, E-views software is useful for time series analysis. However, other applications like STATA and SPSS can only generate the missing values for the data.
Sometimes, the database requires high-frequency data which require hours and seconds frequencies. In addition to this, there is a multi-year, bimonthly, fortnight, ten day and daily frequency. Such type of data is supported using the E-views package with a big size database.
Database often requires a long run variance and covariance calculations within which the errors are distributed randomly and indicate a particular type of sequence. E-views allows calculating symmetric or one-sided long-run variances using the nonparametric kernel. Apart from this, there are pre whitened kernel and parametric VARHAC methods. Furthermore, E-views also support automatic bandwidth selection methods for the kernel estimators and the information criteria which are based upon lag length.
Time series and estimation based disadvantage of E-views
Even though E-views package is useful for the time series analysis, it is not useful for conducting panel data analysis with long term data sets. This is because E-views package is restricted with the matrix of a number of observations and variables. So, obtaining the results for panel data sets may not be feasible.
The algorithm used in E-views for the estimation of the parameters within the log likelihood may not be suited to obtain the results for arbitrary maximization or minimization. This is because the algorithms are based upon the sum of the outer product of the derivatives of likelihood contribution. Due to this, obtaining good approximation with the general setting may not be possible. Specification of the functional form and the statistical properties is required. They are used for maximum likelihood estimation.
Matrix operation and estimation based challenges
Matrix operations in the likelihood estimation in E-views is not possible. To specify the likelihood functions for different statistical models, write the expressions. Additionally, specify the determinants and the quadrant form for each model. This becomes a tedious task while working with two or more than two models.
Estimation in E-views does not optimize the regression model subject to the inequality constraints. However, there exist multiple techniques for the imposition of the inequality constraint in the models. Though this problem can be solved using the monotonic transformation of the mode, there are one-sided or two-sided restrictions. These restrictions may affect the efficiency of the results.
Basic analysis disadvantages
The transformation approach used for optimization is limited to solve simple inequality constraints. Additionally, in case of the cross coefficient inequality restrictions, the solutions can’t be tracked. On the other hand, it is much easier to impose restrictions on statistical packages like STATA.
Performing a hypothesis test in E-views using the untransformed coefficient is also challenging. It requires the estimates for each of the associated expression. Additionally, the variances manually have to be obtained by using the data method. This is due to the fact that the transformations are non-linear. Further, E-views may provide inequality restrictions for the values near the boundaries.
- Aljandali, A. and Tatahi, M. (2018) ‘Introduction to EViews’, in. doi: 10.1007/978-3-319-92985-9_1.