Econometric Modeling of GDP Time Series.docVIP

Theoretical and Applied Economics Volume XVIII (2011), No. 10(563), pp. 91-98 Econometric Modeling of GDP Time Series Elena-Adriana ANDREI Bucharest Academy of Economic Studies adrianna_andrei@ Elena BUGUDUI “Artifex” University of Bucharest bugudui@ Abstract. Article aims of time series econometric model of macroeconomic variable GDP in the US economy. Because that is a nonstationary time series, there are used several statistical tests in order to turn into a stationary series. After applying these tests, the time series became stationary and integrated of order I; thus, we use Box-Jenkins procedure for the determination of ARMA. We estimate by OLS the parameters of various models. Performances chosen ARIMA model (1,1,1) are verified on the basis of classical statistical tests and forecasting. Keywords: stationary time series; nonstationary time series; statistical tests. JEL Codes: C22, E01. REL Codes: 8C, 10G.  92 Elena-Adriana Andrei, Elena Bugudui In order to set time series, we use macroeconomic variable GDP during 1947 (first quarter) – 2010 (third quarter) in the US economy. Times series values are observed at the same frequency, namely quarterly; each of these values are random variables. Therefore, we can say that GDP is a stochastic process and the actual values observed in period mentioned above are individual achievements of this process. Descriptive analysis time series gives us informations about the following indicators: i) time series mean: T ∑ yt y = t=1 = 6,372.390 ? T ii) time series variance: 2 T ? ? T ?1 ? ? ∑ ? ? yt y? s 2 = t=1 ? = 12,735,711 iii) Jarque-Berra statistic value, which suggest a normal distribution of time series in terms of asymetry and flattening. 28 Series: Y Sample 1947:1 2010:3 Observations 255 24 20 16 12 8 Mean Median 6372.390 5776.600 1336