Sankhya: The Indian Journal of Statistics

1996, Volume 58, Series B, Pt. 2, pp. 288--301

BREAKING TREND FUNCTION IN MACRO VARIABLES : THE CASE OF INDIA : 1900--1988

By

ANITA GHATAK, De Montford University

SUMMARY.  This paper tests the hypothesis of difference stationarity of macroeconomic timeseries against the alternative of trend-stationarity with and without allowing for possible structural breaks. The methodologies used are that of Dicky and Fuller (1979) familiarised by Nelson and Plosser (1982) and of dummy variables familiarised by Perron (1989). The country chosen for this study is India during  the period 1900-1988 for twelve macroeconomic variables and the breakpoints chosen are 1914 and 1939 due to the two World Wars, 1947 due to the political Independence and 1951 due to the introduction of economic planning in India. The conventional Dicky-Fuller methodology without allowing for structural breaks cannot reject the hypothesis of difference stationarity for any series. The years 1947 and 1951 made significant changes in the rates of growth of seven out of twelve series. Allowing for break in the level of the series leads to the rejection of the hypothesis of difference stationarity for three series after the year of independence.

AMS (1991) subject classification.  C22.

Key words and phrases. Difference Stationarity, trend stationarity, structural breaks.

Full paper (PDF)