I have daily data for six stock market indexes over the period 2009-2018. However and due to holiday impact the No of observation are different. Is it possible to measure the spillover effect in this case ?
No! Is not possible to measure spillover effect if there is no synchronization of data. Again, many of the programming software (like RATS) will simply not function or "run" because of the missing values or unequal No of observations.
So, you can use any of the methods (particularly the one recommended by Neaime (2012)) of filing missing observations arising from holiday and special events, to fill the missing values before further analysis.