Q) What is the best way to match CRSP and TAQ data? I tried using CRSP TICKER variable, but it does not seem to correspond to TAQ SYMBOL variable. For instance, for ROYAL DUTCH SHELL PLC, CRSP ticker is RDS, but TAQ's symbols are RDSA and RDSB. What would you recommend to use to make sure that the link is accurate?
A) The stock symbol in TAQ is not the same as a ticker in CRSP. CRSP ticker is not always the real exchange ticker of the company, it is cleaned by CRSP according to certain CRSP filters.
CRSP carries a variable called TSYMBOL (see dsenames table in CRSP directory on the WRDS server for the historical map between CRSP TICKER and TSYMBOL). TSYMBOL is the real exchange ticker.
For instance, for your example, you would see that CRSP ticker RDS corresponds to TSYMBOL RDSB and RDSA (two share classes of ROYAL DUTCH SHELL PLC stock). CRSP would separate these share classes in the cusip they provide, but not in CRSP Ticker variable, which would be the same (RDS).
Therefore, in order to get an accurate match between TAQ and CRSP, one needs to use, first, historical cusips (NCUSIP in CRSP and CUSIP in TAQ Master files), and, second, for those stocks that remain unmatched, use TSYMBOL in CRSP and SYMBOL in TAQ with date conditions (because tickers can be reused).