Long memory in the Portuguese stock market

Purpose This paper's aim is to test for the presence of fractional integration, or long memory, in the daily returns of the Portuguese stock market using autoregressive fractionally integrated moving average (ARFIMA), generalised autoregressive conditional heteroskedasticity (GARCH) and ARFIMA-FIGARCH models. Design/methodology/approach The data cover two periods: 4 January 1993-13 January 2006 (full sample), and 1 February 2002-13 January 2006 (that is, data are considered after the merger of the Portuguese Stock Exchange with Euronext). Findings The results from the full sample show strong evidence of long memory in stock returns. When data after the merger are considered, weaker evidence of long memory is found. It is concluded that the Portuguese stock market is more efficient after the merger with Euronext. Originality/value The findings of this paper are helpful to financial managers and investors dealing with Portuguese stock indices.