Econometric Modelling and Forecasting Using EViews Training Course

Participants can prepare datasets and identify data-quality issues before modelling.

• Missing-value codes, interpolation choices and deletion rules for time series
• Outlier screening using plots, z-scores and domain-based investigation
• Descriptive statistics: mean, median, variance, skewness and kurtosis
• Histograms, boxplots and line charts for distribution and trend inspection
• Correlation matrices interpreted without confusing association and causation
• Transforming nominal values into real values using price deflators

Participants can apply transformations that align variables with model assumptions and business questions.

• Logarithmic transformations for elasticities, scale effects and variance stabilisation
• First differences and growth rates calculated from level series
• Seasonal adjustment concepts and calendar effects in monthly data
• Index numbers rebased for comparable price, volume and activity measures
• Lagged variables constructed with explicit economic timing assumptions
• Ratios and per-capita measures for meaningful cross-period comparison

Participants can estimate and interpret a multiple linear regression model in EViews.

• Ordinary least squares assumptions and coefficient estimation logic
• Dependent, explanatory and control variables specified from economic theory
• Regression equations entered using EViews command and object interfaces
• Coefficient signs, magnitudes and units interpreted for decision questions
• R-squared, adjusted R-squared and standard error of regression
• Residuals examined as the basis for model diagnostic testing

Participants can test regression hypotheses and report statistically defensible conclusions.

• Null and alternative hypotheses written for parameter restrictions
• t-tests for individual regression coefficients and directional hypotheses
• F-tests for joint restrictions and nested model comparisons
• Confidence intervals interpreted alongside point estimates and uncertainty
• p-values used without treating arbitrary thresholds as proof
• Type I and Type II errors in management decisions

Participants can specify nonlinear relationships and categorical effects within regression models.

• Level-level, log-level, level-log and log-log regression specifications
• Elasticity and semi-elasticity interpretation from transformed regression coefficients
• Dummy variables representing policy changes, regimes and categories
• Interaction terms testing whether relationships differ between groups
• Piecewise linear specifications for threshold and capacity effects
• Model selection guided by theory, diagnostics and interpretability

Participants can identify specification problems and revise explanatory-variable choices.

• Variance inflation factors diagnosing problematic explanatory-variable correlation
• Pairwise correlations compared with multicollinearity diagnostic evidence
• Omitted-variable bias explained through correlated missing determinants
• Irrelevant regressors assessed through theory and parsimony principles
• Ramsey RESET test for functional-form specification errors
• Restricted and unrestricted models compared using information criteria

Participants can test for unequal error variance and select appropriate corrective methods.

• Heteroskedasticity patterns identified from residual-versus-fitted value plots
• Breusch-Pagan and White tests applied to regression residuals
• Heteroskedasticity-consistent standard errors for valid coefficient inference
• Weighted least squares for known variance structure assumptions
• Log transformations considered when variance rises with variable scale
• Diagnostic findings documented with test statistics and remedial decisions

Participants can detect serial correlation and specify dynamic regression models.

• Autocorrelation diagnosed using residual plots and correlograms
• Durbin-Watson statistic interpreted within appropriate regression conditions
• Breusch-Godfrey LM test for higher-order serial correlation
• Lagged dependent variables introduced with clear economic justification
• Cochrane-Orcutt and Prais-Winsten correction methods compared
• Newey-West standard errors for autocorrelation and heteroskedasticity

Participants can generate conditional forecasts from regression models and explain assumptions.

• Forecast scenarios built from explicit explanatory-variable assumptions
• Static and dynamic forecasts compared for lagged regression models
• Forecast intervals interpreted as uncertainty bands around projections
• Actual-versus-fitted charts used for in-sample performance assessment
• Ex ante and ex post forecasting distinguished for planning use
• Forecast assumptions recorded for governance and management review

Participants can assess time-series properties and choose suitable transformation strategies.

• Trend, seasonality, cycles and irregular variation distinguished visually
• Autocorrelation and partial autocorrelation functions interpreted systematically
• Weak stationarity requirements for standard time-series model estimation
• Unit roots explained through persistent shocks and stochastic trends
• Augmented Dickey-Fuller tests specified with trend and intercept choices
• Phillips-Perron and KPSS tests compared for stationarity evidence

Participants can identify candidate ARIMA models using time-series diagnostics.

• Box-Jenkins workflow: identification, estimation, diagnostics and forecasting
• Autoregressive processes linked to persistence in observed series
• Moving-average processes linked to serially correlated shocks
• ARIMA orders selected from differencing, ACF and PACF patterns
• Seasonal ARIMA terms specified for recurring monthly or quarterly effects
• AIC, Schwarz criterion and Hannan-Quinn criterion compared

Participants can estimate ARIMA models and assess their forecasting performance.

• ARIMA equations estimated using EViews time-series model objects
• Residual correlograms checked for remaining serial correlation patterns
• Ljung-Box Q statistics used for residual white-noise testing
• Holdout samples separated from estimation periods for validation
• MAE, RMSE and MAPE compared across competing forecasts
• Forecast bias assessed through mean error and tracking signals

Participants can estimate lagged relationships and interpret short-run and long-run effects.

• Distributed lag models represent delayed economic response patterns
• Lag length selected using theory, diagnostics and information criteria
• ARDL specifications combine dependent-variable and explanatory-variable lags
• Short-run coefficients separated from long-run equilibrium effects
• Bounds testing assesses level relationships among mixed integration orders
• Dynamic multipliers display adjustment paths following explanatory-variable changes

Participants can test cointegration and estimate error-correction relationships for non-stationary series.

• Cointegration distinguishes shared long-run movement from spurious regression
• Engle-Granger procedure estimates residual-based cointegration relationships
• Johansen tests identify multiple cointegrating vectors in multivariate systems
• Error-correction terms measure speed of return to equilibrium
• Long-run coefficients interpreted separately from short-run dynamics
• Residual diagnostics confirm adequacy of estimated error-correction models

Participants can estimate and interpret VAR models for jointly determined variables.

• VAR models treat relevant variables as jointly endogenous processes
• Lag order selected using information criteria and residual diagnostics
• Stability conditions checked through inverse roots of characteristic polynomial
• Granger causality tests assess predictive content between series
• Impulse response functions trace shocks through dynamic systems
• Forecast error variance decomposition attributes uncertainty across shocks

Participants can select models appropriate to common economic and financial forecasting requirements.

• Inflation forecasts combining price indices, exchange rates and monetary indicators
• GDP nowcasting using monthly production, trade and survey series
• Interest-rate modelling with inflation expectations and policy-rate lags
• Exchange-rate models assessed against random-walk benchmark forecasts
• Credit growth forecasting using income, rates and portfolio indicators
• Corporate sales forecasts linked to macroeconomic and sector drivers

Participants can document, challenge and communicate econometric models for review and use.

• Model purpose, users and decision thresholds documented before deployment
• Assumption registers capture data limits, exclusions and judgement calls
• Backtesting schedules define frequency, benchmarks and escalation thresholds
• Version control records equation changes, data revisions and approvals
• Sensitivity analysis tests material assumptions and alternative specifications
• Technical reports separate statistical findings from decision recommendations

Participants can automate recurring EViews tasks and produce traceable model outputs.

• EViews programs automate imports, transformations, estimation and forecast updates
• Scalar, string and loop commands support repeatable modelling procedures
• Command logs preserved for transparent replication of analytical results
• Spools assemble tables, charts and equations into report outputs
• Named samples control estimation, validation and forecast-period windows
• Automated output checks flag missing observations and failed estimations

Participants can communicate forecast results, uncertainty and model limitations to decision-makers.

• Forecast narratives distinguish evidence, assumptions, judgements and recommendations
• Fan charts communicate central projections and uncertainty ranges
• Scenario tables show consequences of alternative macroeconomic assumptions
• Executive summaries state material drivers, risks and decision implications
• Model limitations communicated without overstating statistical certainty
• Review meetings structured around variance explanations and forecast actions

Participants can build, review and present a complete EViews econometric forecasting study with documented evidence and recommendations.

• Business question translated into measurable outcome and forecast horizon
• EViews workfile assembled from documented source data and transformations
• Candidate regression and time-series models estimated against stated criteria
• Diagnostics, stability tests and residual evidence reviewed systematically
• Out-of-sample forecast comparison selects a defensible final specification
• Management presentation explains assumptions, results, risks and recommended actions