I am currently conducting an event study, in which I want to estimate the impact of a dividend event on trading volumes. I have 84 different companies (with more than one event per company), which all have their own trading volumes (obviously). Therefore I am struggling with how to conduct my final regression.
Short data example for only two companies with one event for each company:
Code:
* Example generated by -dataex-. To install: ssc install dataex clear input float co_id str19 security float date str8 year double(unadjusted_price adjusted_price volume) int dividend_date byte tax float(days_co estimationwindow eventwindow) double tw float lnvol 1 "1&1 DRILLISCH" 17671 "19052008" 4.64 4.64 84.2 0 . 100 1 1 1 4.433195 1 "1&1 DRILLISCH" 17672 "20052008" 4.57 4.57 81.5 0 . 101 1 1 1 4.400603 1 "1&1 DRILLISCH" 17673 "21052008" 4.53 4.53 30.6 0 . 102 1 1 1 3.421 1 "1&1 DRILLISCH" 17674 "22052008" 4.68 4.68 27.7 0 . 103 1 1 1 3.321432 1 "1&1 DRILLISCH" 17675 "23052008" 4.29 4.29 93.5 0 . 104 1 1 1 4.5379615 1 "1&1 DRILLISCH" 17678 "26052008" 4.27 4.27 61.3 0 . 105 1 1 1 4.11578 1 "1&1 DRILLISCH" 17679 "27052008" 4.47 4.47 36.5 0 . 106 1 1 1 3.597312 1 "1&1 DRILLISCH" 17680 "28052008" 4.45 4.45 19.8 0 . 107 1 1 1 2.985682 1 "1&1 DRILLISCH" 17681 "29052008" 4.46 4.46 12.7 0 . 108 1 1 1 2.541602 1 "1&1 DRILLISCH" 17682 "30052008" 4.32 4.32 34.6 0 . 109 1 1 1 3.543854 1 "1&1 DRILLISCH" 17685 "02062008" 4.4 4.4 4.3 1 1 110 1 1 1 1.458615 1 "1&1 DRILLISCH" 17686 "03062008" 4.23 4.23 70.3 0 . 111 1 1 1 4.252772 1 "1&1 DRILLISCH" 17687 "04062008" 4.26 4.26 68.3 0 . 112 1 1 1 4.22391 1 "1&1 DRILLISCH" 17688 "05062008" 4.44 4.44 41.5 0 . 113 1 1 1 3.7256935 1 "1&1 DRILLISCH" 17689 "06062008" 4.4 4.4 76.6 0 . 114 1 1 1 4.3385973 1 "1&1 DRILLISCH" 17692 "09062008" 4.3 4.3 41.5 0 . 115 1 1 1 3.7256935 1 "1&1 DRILLISCH" 17693 "10062008" 4.17 4.17 72.9 0 . 116 1 1 1 4.2890887 1 "1&1 DRILLISCH" 17694 "11062008" 4.05 4.05 72.9 0 . 117 1 1 1 4.2890887 1 "1&1 DRILLISCH" 17695 "12062008" 4.08 4.08 23.5 0 . 118 1 1 1 3.157 1 "1&1 DRILLISCH" 17696 "13062008" 4.16 4.16 127 0 . 119 1 1 1 4.844187 1 "1&1 DRILLISCH" 17699 "16062008" 4.15 4.15 30 0 . 120 1 1 1 3.4011974 2 "AAREAL BANK" 17671 "19052008" 22.25 20.023 357.2 0 . 100 1 1 1 5.878296 2 "AAREAL BANK" 17672 "20052008" 21.6 19.438 635.6 0 . 101 1 1 1 6.454569 2 "AAREAL BANK" 17673 "21052008" 21.42 19.276 426.4 0 . 102 1 1 1 6.055378 2 "AAREAL BANK" 17674 "22052008" 20.94 18.845 236.3 1 1 103 1 1 1 5.465102 2 "AAREAL BANK" 17675 "23052008" 20.01 18.008 647.9 0 . 104 1 1 1 6.473736 2 "AAREAL BANK" 17678 "26052008" 19.9 17.909 138.2 0 . 105 1 1 1 4.928702 2 "AAREAL BANK" 17679 "27052008" 19.35 17.414 461.2 0 . 106 1 1 1 6.133832 2 "AAREAL BANK" 17680 "28052008" 19.37 17.432 255.5 0 . 107 1 1 1 5.543222 2 "AAREAL BANK" 17681 "29052008" 19.08 17.171 381.8 0 . 108 1 1 1 5.944897 2 "AAREAL BANK" 17682 "30052008" 19.13 17.216 318.8 0 . 109 1 1 1 5.764564 2 "AAREAL BANK" 17685 "02062008" 18.77 16.892 287.1 0 . 110 1 1 1 5.659831 2 "AAREAL BANK" 17686 "03062008" 18.92 17.027 164.8 0 . 111 1 1 1 5.104733 2 "AAREAL BANK" 17687 "04062008" 18.82 16.937 224.3 0 . 112 1 1 1 5.412984 2 "AAREAL BANK" 17688 "05062008" 19.01 17.108 306.7 0 . 113 1 1 1 5.72587 2 "AAREAL BANK" 17689 "06062008" 18.87 16.982 331.4 0 . 114 1 0 1 5.803326 2 "AAREAL BANK" 17692 "09062008" 18.46 16.613 421 0 . 115 1 0 1 6.042633 2 "AAREAL BANK" 17693 "10062008" 18.18 16.361 194.8 0 . 116 1 0 1 5.271974 2 "AAREAL BANK" 17694 "11062008" 17.59 15.83 318.2 0 . 117 1 0 1 5.76268 2 "AAREAL BANK" 17695 "12062008" 17.42 15.677 807.8 0 . 118 1 0 1 6.694314 2 "AAREAL BANK" 17696 "13062008" 17.61 15.848 331.3 0 . 119 1 0 1 5.803024 2 "AAREAL BANK" 17699 "16062008" 17.72 15.947 221 0 . 120 1 0 1 5.398163 end format %td date
Code:
sort co_id summ co_id local total_co = `r(max)' forvalues i = 1/`total_co' { xtmixed lnvol dividend_date eventwindow tw if co_id == `i' & estimationwindow==1 }
However, this command will result in 84 outputs (two in the data example), one for each company. My question is how to average these outputs into one output? Or if there is a command which accounts for each company individually but delivers just one average output.
Thank you very much in advance for your help!
0 Response to Combining n regression outputs to one average output
Post a Comment