Is There a Cointegration Amongst Sector-Specific Indices of the Indian Stock Markets? An Investigation Using ARDL with Single Structural Break

The present study makes an attempt to investigate the comovement and cointegration amongst the four sector-specific indices of the National Stock Exchange (NSE), viz., Nifty Auto, Nifty FMCG, Nifty Pharma and Nifty Realty for the ten-year period April 2009 to March 2019. The data, for the purpose of the study, includes monthly returns of these four sector-specific indices. The methodology employed is Autoregressive Distributed Lag (ARDL) bounds testing cointegration approach after incorporation of one break period. The study also tests the necessary prerequisites for the ARDL model, which includes the variable stationarity (using Breakpoint Unit Root test), stability of parameters (using CUSUM plots) and serial correlation (using BG-LM test). The results could not directly identify any cointegration amongst the sampled indices. No inference about cointegration (neither accept nor reject) could be drawn in three of the possible four cases, i.e., Nifty Pharma, Nifty Realty and Nifty FMCG, as their computed values were falling within the upper and lower bound critical limits of ‘F’ bounds test. The results of the fourth index, Nifty Auto, however, showed no cointegration. Since the result of three indices did not give a clear direction about cointegration, it was decided to establish an Error Corrective Mechanism (ECM) for each of these three indices. The results of the error correction showed that for all the three indices, the ECM term was found to be significant and negative. The results of stationarity revealed that Nifty Auto, Nifty FMCG and Nifty Pharma are stationary at first difference, while Nifty Realty was stationary at levels, hence the need for ARDL cointegration model. Also, none of the four-sampled indices suffered from autocorrelation