I got monthly data of the standard deviation of my betas (Sd_beta) and three dummy variables P1-P3, which indicate wheter the volatility of Ted (Ted_Vol) is in the first second or third tercile.
I've regressed this dummies on the above mentioned standard Deviation and got the coefficients for P1 and P3. In the next step I have to determine if the difference (P3-P1) between the standard deviation given that P1=1 or P3 is statistically significant. I'm not qute sure how to approach this task. Is there a way to compute monthly differences and use them for a t-test?
Another thought of mine was to calculate the difference of Ted_Vol if P1=1 and P3=1 but I do not know how to match this numbers regarding my time variable.
Code:
* Example generated by -dataex-. To install: ssc install dataex clear input float date int Jahr byte Monat double(Sd_Beta Ted_Vol) float(P1 P3) 469 1999 2 .2162872850894928 .00042292120633646846 1 0 470 1999 3 .21683630347251892 .0005120532005093992 1 0 471 1999 4 .21574001014232636 .0005089303012937307 1 0 472 1999 5 .21684658527374268 .001667482778429985 0 0 473 1999 6 .21839885413646698 .0011676736176013947 0 0 474 1999 7 .21961617469787598 .0005640562158077955 0 0 475 1999 8 .22290168702602386 .0003095806168857962 1 0 476 1999 9 .23819047212600708 .001692846417427063 0 1 477 1999 10 .2405623197555542 .001672371756285429 0 0 478 1999 11 .23849868774414063 .0029762284830212593 0 1 479 1999 12 .24061444401741028 .002350094262510538 0 1 480 2000 1 .23947422206401825 .0033114443067461252 0 1 481 2000 2 .23897820711135864 .0017416990594938397 0 1 482 2000 3 .23724332451820374 .0005756043246947229 0 0 483 2000 4 .23789244890213013 .0004283250018488616 1 0 484 2000 5 .23350174725055695 .001657421002164483 0 1 485 2000 6 .23567992448806763 .002668452449142933 0 1 486 2000 7 .23621943593025208 .0005725464434362948 1 0 487 2000 8 .23868688941001892 .0008709787507541478 0 0 488 2000 9 .24259942770004272 .0012692536693066359 0 0 489 2000 10 .24906545877456665 .00048570294165983796 1 0 end format %tm date
Thank you in advance.
0 Response to ttest for time series
Post a Comment