European Journal of Business and Management www.iiste.org ISSN 2222-1905 (Paper) ISSN 2222-2839 (Online) Vol 4, No.5, 2012 40 On Structural Breaks and Nonstationary Fractional Intergration in Time Series Olanrewaju I. Shittu 1 OlaOluwa S. Yaya 2* Raphael A. Yemitan 3 123 Department of Statistics, University of Ibadan, Nigeria * E-mail of the corresponding author: os.yaya@ui.edu.ng Abstract The growth of an economy is determined largely by the growth of its Gross Domestic Product (GDP) over time. However, GDP and some economic series are characterized by nonstationarity, structural breaks and outliers. Many attempts have been made to analyze these economic series assuming unit root process even in the presence of changes in the mean level without considering possible fractional integration. This paper aims at examining the structural breaks and nonstationarity in the GDP series of some selected African countries with a view to determining the influence of structural breaks on the level of stationarity of these series. These series are found to be nonstationary with some evidence of long memory. They were found to experience one or more breaks over the years and this may be due to instability in the government and economic policies in the selected African countries. The measure of relative efficiency shows that autoregressive fractional integrated moving average (ARFIMA) models is better than the corresponding autoregressive integrated moving average (ARIMA) models for the series considered in this study. Keywords: fractional integration, gross domestic product, structural breaks 1. Introduction Economic growth for many countries is majorly determined by the country’s Gross Domestic Product (GDP). Among African countries South Africa is rated as the richest country because of her highest value of GDP each year. For this reason, it is sensible to study the pattern in which this is realized over the years bearing in mind that the series are usually nonstationary. Most researches in economic time series have concentrated on the behavior of other economic measures and model are fitted to the series but fewer articles have considered GDP. Economic and financial time series often display properties such as breaks, heteroscedasticity, missing values, outliers, nonlinearity just to mention a few. Of much importance in time series is the structural break or mean shift which affect the level of stationarity in the series. Quite a number of articles have shown that break in structure of the series may cause a stationary series (29 0 I to be fractionally integrated (Granger and Hyung, 2004; Ohanissian et al., 2008). In the context of nonstationary series, there are fewer articles to show the effect of breaks in the series. Chivillon (2004) in the discussion paper on “A