# Fama and French data: Replicating research

Is there any data out there on the Fama and French (1993) paper? I am not talking about their factor data available on their website, I am interested in reproducing their factor calculations. I am trying to replicate their research, such as the SMB, HML calculations given on their website.

I have requested access to WRDS which would give me access to the CRSP and Compustat data, then I believe that I should be fine to replicate their research but I was just wondering if the returns data from NYSE, AMEX and NASDAQ from July 1963 - Dec 1991 was available online.

Example of the data:

        Mkt-RF     SMB     HML      RF
199007    0.86    0.77   -0.25    0.68
199008  -10.82   -1.60    0.60    0.66
199009  -11.97    1.22    0.80    0.60
199010    9.56   -7.38   -4.24    0.68


I would like to calculate myself the SMB and HML factors.

Your best shot is really to get WRDS. If you get access you can replicate their research in few minutes (there is SAS code out there). It's unlikely that you will be given the data for free. That data is actually expensive. That being said, any university with an econ of finance department has access to that data and gives it to students for free. So if you are enrolled in any sort of higher education program you should manage to get access to it fairly easily.

• Thanks for the reply, I am currently doing a PhD so I should be able to get WRDS access but I was hoping to get some preliminary data or FF reproducible data from their website or research paper. – user113156 Nov 18 '17 at 23:13
• No. Not exactly. That would breach the data policy of both csrp and compustat. – phdstudent Nov 18 '17 at 23:18
• That is what I was afraid of. I will await further reply from my application to WRDS. – user113156 Nov 19 '17 at 0:40
• (+1) Quick comments: (1) Though I agree with most everything you write, "a few minutes" is hyperbole for someone's first time; simply figuring out how to run SAS scripts on wrds can take hours if one has never done it before. (2) If you're writing code yourself, there are a few wrinkles; it's not altogether trivial. (eg. what do FF exactly mean by "good shares and price data at the beginning of $t$?")? (3) FF are quite precise in their language and research. It's a nice reproducing exercise that can be done in about a day (longer or less depending on your familiarity with various systems). – Matthew Gunn Nov 20 '17 at 19:31

Their factor construction comes from data from the Chicago Booth Center for Research on Security Prices:

http://www.crsp.com/products/software-access-tools