The Superiority of Panel SPSM-KSS Fourier Univariate Unit Root Test towards Problematic PFI Time Series Data

Abstract: This paper aims at investigating the properties of problematic time series data with outliers and missing values problems by applying the Sequential Panel Selection Method (SPSM) using Panel KSS unit root test with a Fourier Function. The problematic time series data refer to the real-industrial-data which comprise of the Malaysian construction materials price indices monthly data from January 1980 to December 2013, with base 100 in year 1980 covering four states of Malaysian Peninsular central region; Wilayah Persekutuan Kuala Lumpur, Selangor, Melaka and Negeri Sembilan. The method used in this study is powerful to classify the whole panel using structural breaks as well as nonlinearity control, and determines which series in the panel are stationary processes. The empirical results found that the series of Aggregates, Sand and Roof Materials price indices are all stationary even though there exist different severity of outliers problem and interpolated missing values in the data. The missing values interpolation techniques with respect to this study are nearest neighbor, linear, piecewise cubic spline, shape-preserving piecewise cubic, and their significance based on bootstrap p-values are also shown in this paper. This initial test is important to be considered before any further attempts of time series or forecast modeling can be implemented on the data. The findings are important for the policy makers, contractors as well as subcontractors to further forecast the future prices of construction materials and soon assist them in tender bidding before any agreements on construction projects are made.