Abstract: Based on semiparametric methods,the aggregation effect of long memory property in stock markets volatility has been researched from two aspects:On one hand,stock returns are aggregated firstly, then the long memory property of aggregated series is considered;on the other hand,stock returns are convened to volatility series firstly,then the long memory property of aggregated volatility series is considered.The former considers the aggregated property of long memory parameter in volatility series,whilte the latter studies the influence of data frequency on long memory parameter in volatility.Starting from the real situation of Chinese stock markets andintegrated consideration of the two facts,the aggregation invariant effect of long memory property is found in volatility,which can also show the good effect of semiparametric methods.

Key words: volatility, long memory property, aggregated effect, high frequency