Is there Long Memory in Stock Markets, or Does it Depend on the Model, Period or Frequency?

Abstract

This paper analyses the existence of long memory in the major stock markets in the world, and if this is the case, whether it’s due to the type of econometric models used, the period of study or the frequency of data (intraday, daily, weekly, etc.)? To do this, we perform a comparative analysis between the empirical results of ARFIMA and GARCH models. The stock markets that showed consistent results of long memory, regardless of the method, the period and the frequency were China and South Korea. The first one exhibits long memory, and the other a short one.

Keywords

• Mercados bursátiles
• Memoria larga
• Métodos econométricos de series de tiempo