Model uncertainty and forecast accuracy

Article Abstract:

Forecast accuracy in a time-series analysis depends on the accuracy of the model. Time-series data presumes the existence of a model which may not accurately reflect the parameters for forecast conditions. The least square theory, for example, is inappropriate if the data is used to formulate and fit the model, since model accuracy can be compromised with the assumption of incorrect prediction intervals. Forecasting can be improved by Bayesian model averaging and basing the forecast not only on the model.

Statistical hypothesis testing

---

Forecasting with preliminary data

Article Abstract:

Addition of preliminary data is shown to enhance forecast performance using revised American trade balance statistics. Evaluation of benchmark forecasts revealed that an approach that utilizes optimal combination was better compared to an approach that depends entirely on revised data. On the other hand, a bivariate AR error-correction model with GARCH effects yielded better forecasting results compared with other benchmark forecasts that uses preliminary data in the stage of forecast combinations.

Economic forecasting

---

A comparison between linear and nonlinear forecasts for nonlinear AR models

Article Abstract:

An evaluation of linear and nonlinear forecasts for nonlinear AR models was conducted to assess their relative forecast performance. The evaluation was guided by a conditional probability premise that the nonlinear forecast's absolute forecast error is smaller compared to the linear forecast's. A new forecast was formulated based on a finding that the forecast performance of a nonlinear model is dependent to its forecast origins.

Analysis, Prediction theory

Similar abstracts:

• Abstracts: Currency forecasting based on an error components-seemingly unrelated nonlinear regression model. Non-linear, non-parametric, non-fundamental exchange rate forecasting
• Abstracts: Comparing the accuracy of density forecasts from competing models. Robust evaluation of fixed-event forecast rationality
• Abstracts: Vector autoregression modelling and forecasting. Business cycle analysis and forecasting with a structural vector autoregression model for Wales
• Abstracts: Loss aversion and adaptation in the labor market: empirical indifference functions and labor supply. Category versus continuous survey responses in economic modelling: Monte Carlo and empirical evidence

This website is not affiliated with document authors or copyright owners. This page is provided for informational purposes only. Unintentional errors are possible.