Event Study help

[Arthur Jones]

Hi

i'm doing a 'market model' style event study. I'm looking at the effects of a Fed announcement upon a currency.

As i have no market returns to compute abnormal returns, I will be calculating the theoretical variation in the currency over the preceding 300 days before the event. i will use the mean variation during the estimation window. Theoretical variation (over 300 days, estimation window) = SUM (t=300) [ Variation ]/300.

I have currently undertaken the following steps in preparation for my event study (from do file):
gen ret = ln(px_last[_n+1] / px_last[_n])
destring date, ignore("/") replace
gen day_cnt = _n
gen target_day = day_cnt if date==20092017
egen max_target_day = max(target_day)
gen evday = day_cnt-max_target_day
drop day_cnt target_day
max_target_day
sort evday
gen evt_window=1 if evday>=-10 & evday<=10
gen est_window=1 if evday<=302 & evday>=-2

(px_last is the exchange rate on that date)

(I'm a stata newbie, but i believe that 'gen window floor (day_cnt/300)' would be a command i may use. but i dont know how to get further.)

Thank you

[Nick Cox]

Cross-posted at https://stats.stackexchange.com/ques...-exchange-rate

Please note our policy on cross-posting, which is that you are asked to tell us about it.

There is no data example here, but your handling of dates can, I surmise, only cause problems. If you start with string dates such as

"20/09/2017"

then applying destring to such dates can only produce nonsense. Consider for example

"10/10/2017" 

as another date. After stripping out the slashes and forcing to integers these dates become

10102017
20092017

and are no longer even in the right order. They differ also by nearly 10 million (days?), or about 27000 years.

The way to handle such dates is through the daily() or date() function, as illustrated by

. display %td daily("20/09/2017", "DMY")
20sep2017

For explanation, see

help datetime

That's just one detail, but an important one. Otherwise it's hard to say whether you are on the right lines without a data example.

[Nick Cox]

Thanks for your data example. Did you read https://www.statalist.org/forums/help#stata on how to use dataex to do it? Here is your example again, with the correct way to produce numeric dates added at the end.

I am bailing out now, as event studies in economics are outside my field, but others may be able to help.

* Example generated by -dataex-. To install: ssc install dataex
clear
input str10 date float px_last
"29/12/2017" 3.3125
"28/12/2017" 3.3127
"27/12/2017" 3.3166
"26/12/2017" 3.3092
"22/12/2017" 3.3376
"21/12/2017" 3.3066
"20/12/2017" 3.2961
"19/12/2017" 3.2911
"18/12/2017" 3.2889
"15/12/2017" 3.3024
"14/12/2017" 3.3353
"13/12/2017" 3.3174
"12/12/2017" 3.3192
"11/12/2017" 3.3038
"08/12/2017" 3.2935
"07/12/2017" 3.2933
"06/12/2017" 3.2353
"05/12/2017" 3.2429
"04/12/2017" 3.2454
"01/12/2017" 3.2578
"30/11/2017" 3.2681
"29/11/2017"  3.244
"28/11/2017" 3.2135
"27/11/2017" 3.2262
"24/11/2017" 3.2322
"23/11/2017" 3.2225
"22/11/2017" 3.2248
"21/11/2017" 3.2541
"20/11/2017" 3.2574
"17/11/2017" 3.2593
"16/11/2017" 3.2775
"14/11/2017" 3.3152
"13/11/2017" 3.2794
"10/11/2017" 3.2837
"09/11/2017" 3.2505
"08/11/2017" 3.2505
"07/11/2017" 3.2729
"06/11/2017" 3.2519
"03/11/2017" 3.3138
"01/11/2017"  3.266
"31/10/2017" 3.2713
"30/10/2017" 3.2805
"27/10/2017" 3.2356
"26/10/2017" 3.2973
"25/10/2017" 3.2348
"24/10/2017" 3.2474
"23/10/2017" 3.2354
"20/10/2017" 3.1938
"19/10/2017" 3.1708
"18/10/2017" 3.1707
"17/10/2017" 3.1605
"16/10/2017" 3.1694
"13/10/2017"  3.146
"11/10/2017"  3.173
"10/10/2017" 3.1804
"09/10/2017" 3.1883
"06/10/2017" 3.1555
"05/10/2017" 3.1539
"04/10/2017" 3.1349
"03/10/2017" 3.1438
"02/10/2017" 3.1546
"29/09/2017" 3.1625
"28/09/2017" 3.1829
"27/09/2017" 3.1932
"26/09/2017" 3.1662
"25/09/2017" 3.1597
"22/09/2017"  3.125
"21/09/2017" 3.1382
"20/09/2017" 3.1335
"19/09/2017" 3.1355
"18/09/2017" 3.1376
"15/09/2017" 3.1104
"14/09/2017" 3.1209
"13/09/2017" 3.1353
"12/09/2017" 3.1262
"11/09/2017" 3.1029
"08/09/2017" 3.0872
"06/09/2017" 3.1002
"05/09/2017" 3.1174
"04/09/2017" 3.1393
"01/09/2017" 3.1404
"31/08/2017" 3.1491
"30/08/2017" 3.1596
"29/08/2017" 3.1646
"28/08/2017" 3.1669
"25/08/2017"  3.159
"24/08/2017" 3.1482
"23/08/2017" 3.1424
"22/08/2017"  3.163
"21/08/2017" 3.1651
"18/08/2017" 3.1472
"17/08/2017" 3.1753
"16/08/2017" 3.1524
"15/08/2017" 3.1702
"14/08/2017" 3.1901
"11/08/2017" 3.1937
"10/08/2017"  3.176
"09/08/2017" 3.1557
"08/08/2017" 3.1265
"07/08/2017"  3.126
"04/08/2017" 3.1318
"03/08/2017" 3.1146
"02/08/2017" 3.1137
"01/08/2017" 3.1265
"31/07/2017" 3.1254
"28/07/2017" 3.1316
"27/07/2017" 3.1507
"26/07/2017" 3.1384
"25/07/2017" 3.1725
"24/07/2017" 3.1466
"21/07/2017" 3.1433
"20/07/2017" 3.1213
"19/07/2017" 3.1488
"18/07/2017" 3.1565
"17/07/2017" 3.1824
"14/07/2017" 3.1796
"13/07/2017" 3.2116
"12/07/2017" 3.2084
"11/07/2017" 3.2545
"10/07/2017" 3.2557
"07/07/2017" 3.2811
"06/07/2017" 3.2984
"05/07/2017" 3.2891
"04/07/2017" 3.3096
"03/07/2017" 3.3018
"30/06/2017" 3.3082
"29/06/2017" 3.3029
"28/06/2017" 3.2797
"27/06/2017"  3.314
"26/06/2017" 3.2968
"23/06/2017" 3.3427
"22/06/2017" 3.3422
"21/06/2017" 3.3358
"20/06/2017" 3.3274
"19/06/2017" 3.2828
"16/06/2017" 3.2919
"14/06/2017" 3.2752
"13/06/2017" 3.3144
"12/06/2017" 3.3189
"09/06/2017" 3.2968
"08/06/2017" 3.2615
"07/06/2017" 3.2696
"06/06/2017" 3.2782
"05/06/2017" 3.2966
"02/06/2017" 3.2457
"01/06/2017" 3.2504
"31/05/2017"  3.227
"30/05/2017" 3.2579
"29/05/2017" 3.2569
"26/05/2017" 3.2602
"25/05/2017" 3.2752
"24/05/2017" 3.2776
"23/05/2017" 3.2702
"22/05/2017" 3.2666
"19/05/2017" 3.2543
"18/05/2017" 3.3758
"17/05/2017" 3.1391
"16/05/2017" 3.0968
"15/05/2017" 3.1094
"12/05/2017" 3.1224
"11/05/2017" 3.1399
"10/05/2017" 3.1674
"09/05/2017" 3.1894
"08/05/2017" 3.1981
"05/05/2017" 3.1775
"04/05/2017" 3.1887
"03/05/2017" 3.1662
"02/05/2017" 3.1514
"28/04/2017" 3.1768
"27/04/2017" 3.1846
"26/04/2017" 3.1743
"25/04/2017" 3.1472
"24/04/2017" 3.1279
"20/04/2017" 3.1473
"19/04/2017" 3.1529
"18/04/2017" 3.1081
"17/04/2017" 3.1001
"13/04/2017" 3.1442
"12/04/2017" 3.1256
"11/04/2017" 3.1374
"10/04/2017" 3.1316
"07/04/2017" 3.1464
"06/04/2017" 3.1434
"05/04/2017" 3.1206
"04/04/2017" 3.0935
"03/04/2017" 3.1146
"31/03/2017"  3.122
"30/03/2017" 3.1508
"29/03/2017" 3.1197
"28/03/2017" 3.1413
"27/03/2017" 3.1266
"24/03/2017"  3.108
"23/03/2017" 3.1407
"22/03/2017" 3.0879
"21/03/2017" 3.0879
"20/03/2017" 3.0724
"17/03/2017" 3.0916
"16/03/2017"  3.119
"15/03/2017" 3.1037
"14/03/2017" 3.1691
"13/03/2017" 3.1529
"10/03/2017" 3.1419
"09/03/2017"  3.193
"08/03/2017" 3.1704
"07/03/2017" 3.1193
"06/03/2017" 3.1375
"03/03/2017" 3.1155
"02/03/2017" 3.1534
"01/03/2017" 3.0938
"24/02/2017" 3.1104
"23/02/2017" 3.0631
"22/02/2017" 3.0646
"21/02/2017" 3.0964
"20/02/2017" 3.0878
"17/02/2017" 3.0978
"16/02/2017" 3.0898
"15/02/2017" 3.0574
"14/02/2017" 3.0852
"13/02/2017" 3.1103
"10/02/2017" 3.1151
"09/02/2017" 3.1273
"08/02/2017" 3.1156
"07/02/2017"  3.121
"06/02/2017" 3.1193
"03/02/2017"  3.123
"02/02/2017" 3.1217
"01/02/2017" 3.1281
"31/01/2017" 3.1486
"30/01/2017" 3.1265
"27/01/2017" 3.1409
"26/01/2017" 3.1744
"25/01/2017" 3.1701
"24/01/2017" 3.1705
"23/01/2017" 3.1646
"20/01/2017" 3.1743
"19/01/2017" 3.2018
"18/01/2017" 3.2232
"17/01/2017" 3.2116
"16/01/2017" 3.2414
"13/01/2017" 3.2176
"12/01/2017" 3.1868
"11/01/2017"  3.197
"10/01/2017" 3.1941
"09/01/2017" 3.1997
"06/01/2017" 3.2226
"05/01/2017" 3.1988
"04/01/2017" 3.2213
"03/01/2017"  3.265
"02/01/2017" 3.2859
"30/12/2016" 3.2552
"29/12/2016" 3.2538
"28/12/2016" 3.2778
"27/12/2016" 3.2734
"26/12/2016"   3.28
"23/12/2016" 3.2703
"22/12/2016" 3.2916
"21/12/2016" 3.3309
"20/12/2016" 3.3503
"19/12/2016" 3.3695
"16/12/2016" 3.3879
"15/12/2016" 3.3651
"14/12/2016" 3.3715
"13/12/2016" 3.3344
"12/12/2016" 3.3377
"09/12/2016" 3.3802
"08/12/2016" 3.3753
"07/12/2016" 3.3925
"06/12/2016"   3.41
"05/12/2016"  3.423
"02/12/2016" 3.4769
"01/12/2016" 3.4645
"30/11/2016" 3.3858
"29/11/2016" 3.3928
"28/11/2016" 3.3876
"25/11/2016" 3.4141
"24/11/2016" 3.3968
"23/11/2016" 3.3903
"22/11/2016" 3.3563
"21/11/2016" 3.3516
"18/11/2016" 3.3818
"17/11/2016" 3.4211
"16/11/2016" 3.4262
"14/11/2016" 3.4344
"11/11/2016" 3.4023
"10/11/2016" 3.3932
"09/11/2016" 3.2254
"08/11/2016" 3.1686
"07/11/2016" 3.2061
"04/11/2016" 3.2362
"03/11/2016"  3.242
"01/11/2016" 3.2317
"31/10/2016" 3.1936
"28/10/2016" 3.2006
"27/10/2016" 3.1665
"26/10/2016" 3.1405
"25/10/2016" 3.1127
"24/10/2016" 3.1223
"21/10/2016" 3.1563
"20/10/2016"  3.142
"19/10/2016"  3.169
"18/10/2016" 3.1877
"17/10/2016" 3.2046
"14/10/2016" 3.2062
"13/10/2016" 3.1787
"11/10/2016" 3.1954
"10/10/2016" 3.2065
"07/10/2016" 3.2212
"06/10/2016"  3.228
"05/10/2016" 3.2213
"04/10/2016"  3.258
"03/10/2016" 3.2099
"30/09/2016" 3.2624
"29/09/2016" 3.2597
"28/09/2016" 3.2167
"27/09/2016" 3.2343
"26/09/2016" 3.2424
"23/09/2016"  3.244
"22/09/2016" 3.2206
"21/09/2016" 3.2069
"20/09/2016" 3.2589
"19/09/2016"  3.273
"16/09/2016" 3.2629
"15/09/2016" 3.3026
"14/09/2016" 3.3448
"13/09/2016"  3.314
"12/09/2016" 3.2476
"09/09/2016" 3.2741
"08/09/2016" 3.2143
"06/09/2016" 3.1964
"05/09/2016" 3.2834
"02/09/2016" 3.2569
"01/09/2016" 3.2573
"31/08/2016" 3.2267
"30/08/2016" 3.2401
"29/08/2016" 3.2307
"26/08/2016" 3.2657
"25/08/2016" 3.2345
"24/08/2016" 3.2249
"23/08/2016" 3.2351
"22/08/2016" 3.2007
"19/08/2016"  3.205
"18/08/2016" 3.2392
"17/08/2016" 3.2064
"16/08/2016" 3.2022
"15/08/2016" 3.1864
"12/08/2016" 3.1922
"11/08/2016" 3.1444
"10/08/2016" 3.1274
"09/08/2016" 3.1454
"08/08/2016" 3.1728
"05/08/2016" 3.1655
"04/08/2016" 3.1934
"03/08/2016" 3.2385
"02/08/2016" 3.2591
"01/08/2016" 3.2655
"29/07/2016" 3.2489
"28/07/2016" 3.2918
"27/07/2016" 3.2619
"26/07/2016" 3.2761
"25/07/2016"  3.286
"22/07/2016" 3.2565
"21/07/2016" 3.2719
"20/07/2016"   3.26
"19/07/2016" 3.2498
"18/07/2016" 3.2537
"15/07/2016" 3.2802
"14/07/2016" 3.2526
"13/07/2016" 3.2643
"12/07/2016" 3.2956
"11/07/2016" 3.3094
"08/07/2016"    3.3
"07/07/2016" 3.3675
"06/07/2016"   3.33
"05/07/2016" 3.3033
"04/07/2016" 3.2696
"01/07/2016" 3.2364
"30/06/2016"  3.213
"29/06/2016" 3.2206
"28/06/2016" 3.3032
end

gen ndate = daily(date, "DMY")
format ndate %td

[Nick Cox]

As said, I can't advise how to go further. But I glanced at the rest of the code you gave.

You work in terms of observation numbers _n but your data are only for trading days.

You should consider whether

sort ndate 
gen ret = ln(px_last[_n+1] / px_last)
gen evt_window= inrange(ndate - mdy(9, 20, 2017), -10, 10)
gen est_window= inrange(ndate - mdy(9, 20, 2017), -2, 302)

is closer to what you want (and avoids creating variables you don't need).