About link compustat and crsp by python

[kite8]

I have two questions I would like to ask you. Thanks.

1. How to match CRSP monthly with Compustat quarterly through python?
   Is like this ?

# Get a comp demo data
comp_demo = db.get_table(library='comp', table='funda', columns=['gvkey', 'datadate',
                                                                'fyear'])
comp_demo.drop_duplicates(inplace=True)

# Get a CRSP demo data
crsp_demo = db.get_table(library='crsp', table='msf', columns=['permno', 'ncusip', 'date'])
crsp_demo.drop_duplicates(inplace=True)

# Get Compustat-CRSP linktable
link = db.get_table(library='crsp', table='ccmxpf_lnkhist', columns=['gvkey', 'lpermno', 'lpermco', 'linktype', 'linkprim']) # 'linkdt', 'linkenddt'
link = link[link['linktype'].isin(['LC', 'LU'])]
link = link[link['linkprim'].isin(['C', 'P'])]
link.drop(['linktype', 'linkprim'], axis=1, inplace=True)

crsp_add_gvkey = crsp_demo.merge(link, left_on='permno', right_on='lpermno')
crsp_add_gvkey.drop_duplicates(inplace=True)

out = crsp_add_gvkey.merge(comp_demo, left_on='gvkey', right_on='gvkey')
out.drop_duplicates(inplace=True)

2. how to resample the data by fiscal year ?

[snauhaus]

Hey,

The link tables just help match the correct firms to each other. To match the dates, be it months, quarters, or years, you need to create a matching date variable in both datasets. Then, to aggregate, use whatever statistic makes sense to you. I used the median CRSP forecast as that seemed to be common in the literature.

Hope that helps!

[kite8]

Thank you very much!
I had succeed in linking table.