Font Size: a A A

Alternative Statistical Methods To Forecast China's Gnp

Posted on:2002-11-15Degree:MasterType:Thesis
Country:ChinaCandidate:Ashraf Fetoh EataFull Text:PDF
GTID:2156360062975357Subject:Statistics
Abstract/Summary:PDF Full Text Request
This study deals with developing a statistical model, which could be able to accurately forecast China's GNP values. The researcher depended on many statistical methods (regression methods, time series methods and combine regression-time series methods) to estimate alternative statistical models. And he used several economic, statistic and econometric criteria to decide on the most accurate model for forecasting.The main aim of this study (find out which is the best model to forecast China's GNP) was divided into eight sub-aims: every sub-aim was discussed in separated sections. Section 1. deals with estimating several regression models using explanatory variables and applying suitable diagnostics checking to find the most accurate model. Section 2. deals with estimating a regression model that depends on principal component factors instead of explanatory variables. Section 3 also used regression method with the explanatory variables included on some lagged variables. Section 4. developed a univariate ARIMA method for forecasting China's GNP.Sections 5 to 7 deal with estimating models that depend on combine regression-time series techniques. Section 5 deals with the use of the multivariate time series method. Section 6, developed the model that was selected in section 1 by building an ARIMA model for the residual terms that were generated from the selected model; this ARIMA model was added to the model in section 1, generating a new regression-time series combine model. Section 7, deals with some statistical techniques that could be used in estimating a regression model that has an AR(1) error process.The last section of this practical study (Section 8) deals with the comparison process, which included many criteria to help in selecting the most accurate model of all estimated models in this study.This study depends on the publication data from China Statistical Yearbook issues 1986 and 2000. The researcher used the annual data series from 1952 to 1999 for China's Gross National Product, China's Total exports and China's Total Imports.The work of this study suggests that the multivariate ARIMA (0,1,2) model, which included China's Total Imports as an explanatory variable, is the most accurate model of all estimated models in this thesis.
Keywords/Search Tags:Gross National Product, GNP, China, Forecast, Combine Regression-Time series Methods, Lagged Variables, Univariate ARIMA Model, Multivariate ARIMA model, and Regression with AR(1) error process
PDF Full Text Request
Related items