I'm doing a long-run event study using the methodology of Kolari & Pynnonen (2018), where i need to sum monthly abnormal standardized returns (ASR) over a holding period across firms. I've chosen a 12 month holding period (i.e. roughly 250 trading days, and my data only contains trading days). I have daily data but i have calculated a monthly ASR that is represented in my data for the last day of each month. So what i need to do is first sum ASR for 12 months/250 trading days each time Eventday = 1 within each firm. Next i need to identify the number of event firms across all firms during this period and then aggregate all the ASRs within this period across firms. Lastly, i need to multiply this with a "scaling factor" that you can see in the image at the bottom of this post (n is the number of events in the holding period). To run my t-test i need to identify the standard deviation of ASR (accumulated across firms) during each holding period.
Here is an example of my data:
Code:
* Example generated by -dataex-. To install: ssc install dataex clear input int Date float(mofd month FirmID Eventday ASR) 18266 600 1 1 . . 18267 600 1 1 . . 18268 600 1 1 . . 18269 600 1 1 . . 18270 600 1 1 . . 18273 600 1 1 . . 18274 600 1 1 . . 18275 600 1 1 . . 18276 600 1 1 . . 18277 600 1 1 . . 18281 600 1 1 . . 18282 600 1 1 . . 18283 600 1 1 . . 18284 600 1 1 . . 18287 600 1 1 . . 18288 600 1 1 . . 18289 600 1 1 . . 18290 600 1 1 . . 18291 600 1 1 . 3.065664 18295 601 2 1 . . 18297 601 2 1 . . 18298 601 2 1 . . 18301 601 2 1 . . 18302 601 2 1 . . 18303 601 2 1 . . 18304 601 2 1 . . 18309 601 2 1 . . 18310 601 2 1 . . 18311 601 2 1 . . 18312 601 2 1 . . 18315 601 2 1 . . 18316 601 2 1 . . 18317 601 2 1 . . 18318 601 2 1 . . 18319 601 2 1 1 .13870999 18322 602 3 1 . . 18323 602 3 1 . . 18324 602 3 1 . . 18325 602 3 1 . . 18326 602 3 1 . . 18329 602 3 1 . . 18331 602 3 1 . . 18332 602 3 1 . . 18333 602 3 1 . . 18336 602 3 1 . . 18337 602 3 1 . . 18338 602 3 1 . . 18339 602 3 1 . . 18340 602 3 1 . . 18343 602 3 1 . . 18344 602 3 1 . . 18345 602 3 1 . . 18346 602 3 1 . . 18347 602 3 1 . . 18350 602 3 1 . . 18351 602 3 1 . . 18352 602 3 1 . 1.9073694 18353 603 4 1 . . 18357 603 4 1 . . 18358 603 4 1 . . 18359 603 4 1 . . 18360 603 4 1 . . 18361 603 4 1 . . 18364 603 4 1 . . 18365 603 4 1 . . 18366 603 4 1 . . 18367 603 4 1 . . 18368 603 4 1 . . 18371 603 4 1 . . 18372 603 4 1 . . 18373 603 4 1 . . 18374 603 4 1 . . 18375 603 4 1 . . 18378 603 4 1 . . 18380 603 4 1 . . 18381 603 4 1 . -3.122382 18385 604 5 1 . . 18386 604 5 1 . . 18387 604 5 1 . . 18388 604 5 1 . . 18389 604 5 1 . . 18392 604 5 1 . . 18393 604 5 1 . . 18394 604 5 1 . . 18395 604 5 1 . . 18396 604 5 1 . . 18399 604 5 1 . . 18400 604 5 1 . . 18401 604 5 1 . . 18402 604 5 1 . . 18403 604 5 1 . . 18406 604 5 1 . . 18407 604 5 1 . . 18408 604 5 1 . . 18409 604 5 1 . . 18410 604 5 1 . -1.850041 18414 605 6 1 . . 18415 605 6 1 . . 18416 605 6 1 . . 18417 605 6 1 . . end format %tdnn/dd/CCYY Date format %tm mofd
Question for the econometrics people (if any)
This might be a long shot but below is an image of what i need to do along with the t-test i need to run. I know this isn't an econometrics forum per se, but i suspect that at least some of you are quite familiar with it so i might as well try my luck. It says that if i run an OLS regression with clustering robust standard error i should be able to get "my answer" by looking at the intercept. What i dont understand is that how can i run an OLS regression with only a dependent variable (i.e. ASR) and no independent ones? I could write ttest ASR==0 in STATA, but this doesnt alow me to use the "vce(robust) / vce(cluster)" options that i need in order to have clustering robust standard errors.
Array
0 Response to Summing over a range of observations
Post a Comment