Analyze the stickiness of variable

[Helen Chang]

Hi, I have a firm-year (GVKEY year) panel data. I would like to know how sticky the variable of interest is (i.e., DisclosureDummy, firm discloses information in the year equals one, zero otherwise). I assume if it is very sticky, then adding firm fixed effect in the regression is not appropriate in this setting, correct?
Thus, I want to ask how and what kind of descriptive statistics/analysis I need to look at in order to determine the stickiness of the variable (i.e., DisclosureDummy), and ultimately to see whether firm fixed effect is appropriate.

Thank you in advance!!

* Example generated by -dataex-. To install: ssc install dataex
clear
input str43 CompanyName str6 GVKEY str9 CUSIP double year float(DisclosureDummy DisclosureDummy2) double fyear str16 sic
"ACCO BRANDS"            "164506" "00081T108" 2015 0 0 2015 "2780"
"ACCO BRANDS"            "164506" "00081T108" 2016 0 0 2016 "2780"
"AMAG PHARMACEUTICALS"   "012250" "00163U106" 2016 0 0 2016 "2834"
"AMC NETWORKS CL.A"      "187406" "00164V103" 2011 0 0 2011 "4833"
"AMC NETWORKS CL.A"      "187406" "00164V103" 2012 0 0 2012 "4833"
"AMC NETWORKS CL.A"      "187406" "00164V103" 2013 0 0 2013 "4833"
"AMC NETWORKS CL.A"      "187406" "00164V103" 2014 0 0 2014 "4833"
"AMC NETWORKS CL.A"      "187406" "00164V103" 2015 0 0 2015 "4833"
"AMC NETWORKS CL.A"      "187406" "00164V103" 2016 0 0 2016 "4833"
"AMC ENTERTAINMENT HDG." "177637" "00165C104" 2015 0 0 2015 "7830"
"AMC ENTERTAINMENT HDG." "177637" "00165C104" 2016 0 0 2016 "7830"
"AOL, Inc."              "183920" "00184X105" 2011 0 0 2011 "7370"
"AOL, Inc."              "183920" "00184X105" 2012 0 0 2012 "7370"
"AT&T"                   "009899" "00206R102" 2005 0 1 2005 "4812"
"AT&T"                   "009899" "00206R102" 2006 1 1 2006 "4812"
"AT&T"                   "009899" "00206R102" 2007 0 1 2007 "4812"
"AT&T"                   "009899" "00206R102" 2008 1 1 2008 "4812"
"AT&T"                   "009899" "00206R102" 2009 1 1 2009 "4812"
"AT&T"                   "009899" "00206R102" 2011 1 1 2011 "4812"
"AT&T"                   "009899" "00206R102" 2012 1 1 2012 "4812"
"AT&T"                   "009899" "00206R102" 2013 1 1 2013 "4812"
"AT&T"                   "009899" "00206R102" 2014 1 1 2014 "4812"
"AT&T"                   "009899" "00206R102" 2015 0 1 2015 "4812"
"ABBVIE"                 "016101" "00287Y109" 2013 0 1 2013 "2836"
"ABBVIE"                 "016101" "00287Y109" 2014 1 1 2014 "2836"
"ABBVIE"                 "016101" "00287Y109" 2015 1 1 2015 "2836"
"ABBVIE"                 "016101" "00287Y109" 2016 1 1 2016 "2836"
"ACTIVISION BLIZZARD"    "180405" "00507V109" 2008 0 0 2008 "7372"
"ACTIVISION BLIZZARD"    "180405" "00507V109" 2009 0 0 2009 "7372"
"ACTIVISION BLIZZARD"    "180405" "00507V109" 2010 0 0 2010 "7372"
"ACTIVISION BLIZZARD"    "180405" "00507V109" 2011 0 0 2011 "7372"
"ACTIVISION BLIZZARD"    "180405" "00507V109" 2012 0 0 2012 "7372"
"ACTIVISION BLIZZARD"    "180405" "00507V109" 2013 0 0 2013 "7372"
"ACTIVISION BLIZZARD"    "180405" "00507V109" 2014 0 0 2014 "7372"
"ACTIVISION BLIZZARD"    "180405" "00507V109" 2015 0 0 2015 "7372"
"ACTIVISION BLIZZARD"    "180405" "00507V109" 2016 0 0 2016 "7372"
"ACUITY BRANDS"          "146017" "00508Y102" 2014 0 0 2014 "3640"
"ACUITY BRANDS"          "146017" "00508Y102" 2015 0 0 2015 "3640"
"ACUITY BRANDS"          "146017" "00508Y102" 2016 0 0 2016 "3640"
"ADOBE SYSTEMS"          "012540" "00724F101" 2003 1 1 2003 "7370"
"ADOBE SYSTEMS"          "012540" "00724F101" 2004 1 1 2004 "7370"
"ADOBE SYSTEMS"          "012540" "00724F101" 2005 0 1 2005 "7370"
"ADOBE SYSTEMS"          "012540" "00724F101" 2006 0 1 2006 "7370"
"ADOBE SYSTEMS"          "012540" "00724F101" 2007 1 1 2007 "7370"
"ADOBE SYSTEMS"          "012540" "00724F101" 2008 1 1 2008 "7370"
"ADOBE SYSTEMS"          "012540" "00724F101" 2009 1 1 2009 "7370"
"ADOBE SYSTEMS"          "012540" "00724F101" 2010 1 1 2010 "7370"
"ADOBE SYSTEMS"          "012540" "00724F101" 2011 1 1 2011 "7370"
"ADOBE SYSTEMS"          "012540" "00724F101" 2012 1 1 2012 "7370"
"ADOBE SYSTEMS"          "012540" "00724F101" 2013 1 1 2013 "7370"
"ADOBE SYSTEMS"          "012540" "00724F101" 2014 1 1 2014 "7370"
"ADOBE SYSTEMS"          "012540" "00724F101" 2015 1 1 2015 "7370"
"ADV.AUTO PARTS"         "145977" "00751Y106" 2004 0 0 2004 "5531"
"ADV.AUTO PARTS"         "145977" "00751Y106" 2005 0 0 2005 "5531"
"ADV.AUTO PARTS"         "145977" "00751Y106" 2006 0 0 2006 "5531"
"ADV.AUTO PARTS"         "145977" "00751Y106" 2007 0 0 2007 "5531"
"ADV.AUTO PARTS"         "145977" "00751Y106" 2008 0 0 2008 "5531"
"ADV.AUTO PARTS"         "145977" "00751Y106" 2009 0 0 2009 "5531"
"ADV.AUTO PARTS"         "145977" "00751Y106" 2010 0 0 2010 "5531"
"ADV.AUTO PARTS"         "145977" "00751Y106" 2011 0 0 2011 "5531"
"ADV.AUTO PARTS"         "145977" "00751Y106" 2012 0 0 2012 "5531"
"ADV.AUTO PARTS"         "145977" "00751Y106" 2013 0 0 2013 "5531"
"ADV.AUTO PARTS"         "145977" "00751Y106" 2014 0 0 2014 "5531"
"ADV.AUTO PARTS"         "145977" "00751Y106" 2015 0 0 2015 "5531"
"ADVANCED DRAINAGE SYS." "020126" "00790R104" 2016 0 0 2015 "3080"
"AGILENT TECHS."         "126554" "00846U101" 2003 1 1 2003 "3826"
"AGILENT TECHS."         "126554" "00846U101" 2004 1 1 2004 "3826"
"AGILENT TECHS."         "126554" "00846U101" 2006 1 1 2006 "3826"
"AGILENT TECHS."         "126554" "00846U101" 2007 1 1 2007 "3826"
"AGILENT TECHS."         "126554" "00846U101" 2009 1 1 2009 "3826"
"AGILENT TECHS."         "126554" "00846U101" 2010 1 1 2010 "3826"
"AGILENT TECHS."         "126554" "00846U101" 2011 1 1 2011 "3826"
"AGILENT TECHS."         "126554" "00846U101" 2012 1 1 2012 "3826"
"AGILENT TECHS."         "126554" "00846U101" 2013 0 1 2013 "3826"
"AGILENT TECHS."         "126554" "00846U101" 2014 1 1 2014 "3826"
"AGILENT TECHS."         "126554" "00846U101" 2015 1 1 2015 "3826"
"AKAMAI TECHS."          "125595" "00971T101" 2004 0 1 2004 "7370"
"AKAMAI TECHS."          "125595" "00971T101" 2006 0 1 2006 "7370"
"AKAMAI TECHS."          "125595" "00971T101" 2007 0 1 2007 "7370"
"AKAMAI TECHS."          "125595" "00971T101" 2008 0 1 2008 "7370"
"AKAMAI TECHS."          "125595" "00971T101" 2009 0 1 2009 "7370"
"AKAMAI TECHS."          "125595" "00971T101" 2010 0 1 2010 "7370"
"AKAMAI TECHS."          "125595" "00971T101" 2011 0 1 2011 "7370"
"AKAMAI TECHS."          "125595" "00971T101" 2012 0 1 2012 "7370"
"AKAMAI TECHS."          "125595" "00971T101" 2013 1 1 2013 "7370"
"AKAMAI TECHS."          "125595" "00971T101" 2014 1 1 2014 "7370"
"AKAMAI TECHS."          "125595" "00971T101" 2015 1 1 2015 "7370"
"ALERE"                  "146156" "01449J105" 2010 0 0 2010 "2835"
"ALERE"                  "146156" "01449J105" 2011 0 0 2011 "2835"
"ALERE"                  "146156" "01449J105" 2012 0 0 2012 "2835"
"ALERE"                  "146156" "01449J105" 2013 0 0 2013 "2835"
"ALERE"                  "146156" "01449J105" 2014 0 0 2014 "2835"
"ALERE"                  "146156" "01449J105" 2015 0 0 2015 "2835"
"ALLEGIANT TRAVEL"       "175007" "01748X102" 2015 0 0 2015 "4512"
"ALLY FINANCIAL"         "005072" "02005N100" 2016 0 0 2016 "6172"
"ALTRIA GROUP"           "008543" "02209S103" 2003 0 1 2003 "2111"
"ALTRIA GROUP"           "008543" "02209S103" 2004 0 1 2004 "2111"
"ALTRIA GROUP"           "008543" "02209S103" 2005 0 1 2005 "2111"
"ALTRIA GROUP"           "008543" "02209S103" 2006 0 1 2006 "2111"
"ALTRIA GROUP"           "008543" "02209S103" 2007 0 1 2007 "2111"
end

[Maarten Buis]

You could look at xttab at the within column. Here is the result for your example:

. destring GVKEY, gen(gvkey)
GVKEY: all characters numeric; gvkey generated as long

. xtset gvkey year
       panel variable:  gvkey (unbalanced)
        time variable:  year, 2003 to 2016, but with gaps
                delta:  1 unit

. xttab DisclosureDummy

                  Overall             Between            Within
Disclos~y |    Freq.  Percent      Freq.  Percent        Percent
----------+-----------------------------------------------------
        0 |      66     66.00        18    100.00          80.68
        1 |      34     34.00         5     27.78          69.56
----------+-----------------------------------------------------
    Total |     100    100.00        23    127.78          78.26
                                (n = 18)

If a firm ever chose not to disclose, then it will not disclose 81% of the times (first row, last column). If a firm ever chose to disclose, then will disclose 70% of the times (second row, last column). So this choice is pretty sticky and not disclosing is more sticky then disclosing.

[Helen Chang]

I see. Thank you very much!!