Several papers use measures of delta (pay-performance sensitivity), vega (risktaking incentives), and firm-specific wealth (inside equity) for executives on Execucomp. For example,

1. Core, J., Guay, W., 2002. Estimate the value of employee stock option portfolios and their sensitivities to price and volatility. *Journal of Accounting Research *40, 613-630.

- develops a method to calculate delta and vega using information provided by
*Execucomp*in the pre-2006 period

2. Coles, J., Daniel, N., Naveen, L., 2006. Managerial incentives and risk-taking. *Journal of Financial Economics *79, 431-468.

- estimates delta and vega

3. Daniel, N., Li, Y., and Naveen, L. 2013. No asymmetry in pay for luck. Working Paper.

- calculates a measure of firm-specific wealth using executives’ stock and option portfolios

Naveen makes publicly available her SAS program used to calculate delta, vega, and firm-specific wealth. See her homepage here. However,

- Her program is not self-executable because she uses three external datasets for which she does not provide SAS codes used to create them.
- Her program is to be executed locally so not portable (does not support PC SAS or SSH connection).
- Her program calculates these measures only up to fiscal year 2010.

I improve Naveen’s program to make it self-contained and executable on its own. Specifically, I recreate the three datasets within the new program and update dataset references to point to the sever end. Now you can run the program via PC SAS or SSH connection, and specify the start year and end year of the period of interest. So you can easily update the data up to the most recent date.

I write a little more details in the overview section in the new program. As evidenced in the overview, I believe that I successfully replicate Naveen’s data using the new program. However, if you decide to use the new program, the accuracy of the generated data is your own responsibility.

Lastly, please cite Naveen’s work if you use the new program. I would be appreciated if you are generous enough to acknowledge my work.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 |
/** program reads in data from execcomp and calculates **/ /** delta, vega, and equity portfolio value for all executives **/ %let wrds=wrds-cloud.wharton.upenn.edu 4016; options comamid=TCP remote=WRDS; signon username=_prompt_; libname local 'D:\Dropbox'; options nolabel nocenter; rsubmit; *==================================================================; * OVERVIEW ; *==================================================================; * Naveen's orginal program uses three external datasets. As a result, her program is not executable * on a standalone basis. I create these three datasets within the new program in order to make this * program self-contained and executable through PC/SAS or SSH connection. * Naveen's program generates delta, vega, and firm-specific wealth for the period 1992-2010. * This new program can easily update all measures to the lastest period by specifying the end year. * The three datasets external to Naveen's original program are: * - combined_201111_1_for_wrds * This is a CRSP/Compustat merged dataset containing GVEKY, PERMNO, YEAR (i.e., FYEAR), * FYBEGDT (first day of fiscal year), and FYENDDT (last day of fiscal year, i.e., DATADATE). * - exec_roll_vol_fyear_201111_1 * By Naveen's definition, this is standard deviation of monthly stock returns estimated * over the 60 months prior to the beginning of the fiscal period (i.e., FYBEGDT). * - riskfree_rate * Naveen gets risk-free rates from Fed Reserve website: http://www.federalreserve.gov/releases/h15/data.htm. * These rates can also be accessed on WRDS ("/wrds/frb"). Accessing from WRDS enables * the program conveniently self-executable. * I first recreate the three datasets. For clarity I rename combined_201111_1_for_wrds and * exec_roll_vol_fyear_201111_1 to combined_for_wrds and exec_roll_vol_fyear, respectively. * Naveen makes these measures for the period 1992-2010 available for download. * As a double check, I use this program to recreate these measures for the same period by specifying * start_year=1992 and end_year=2010. The generated data are not exactly the same as Naveen's. * This is not surprising as WRDS continuously updates databases. In fact, the correlation * between the new data and Naveen's data is almost one (delta: 99.99%, vega: 99.93%, and * firm-specific wealth: 99.99%). The double check suggests that this program successfully replicates * Naveen's data. * Other changes I make to Naveen's code include: * * (1) Naveeen add the delta of executive option holdings and the delta of executive stock holdings, and output * the total delta. I also output each component, optiondelta and sharedelta, respectively. * * (2) I also correct an error in Naveen's code. For example, Naveen use the following function: * newdelta = sumSvest_yearend + sumSunvest_yearend + Sshr * This will generate a missing newdelta if any variable on the right hand side is missing, but newdelta * should be missing only when all right-hand-side variables are missing. I correct the error by using the * following function instead: * newdelta = sum(sumSvest_yearend, sumSunvest_yearend, Sshr) *==================================================================; * SPECIFY THE START YEAR AND END YEAR OF THE PERIOD OF INTEREST ; *==================================================================; %let start_year=1992; %let end_year=2014; *==================================================================; * GENERATE THREE DATASETS WHICH ARE EXTERNAL TO NAVEEN'S PROGRAM ; *==================================================================; * Extract Compustat data; data comp1; set comp.funda; where &start_year-6 <= fyear <= &end_year and indfmt='INDL' and datafmt='STD' and popsrc='D' and consol='C'; fyenddt=datadate; fybegdt=intnx('month', fyenddt, -11, 'beg'); format fyenddt fybegdt date9.; keep gvkey datadate fyear fybegdt fyenddt; rename fyear=year; run; proc sort data=comp1; by gvkey datadate; run; * Link GVKEYS to CRSP identifiers; proc sql; create table combined_for_wrds as select a.*, b.lpermno as permno from comp1 as a, crsp.ccmxpf_lnkhist as b where a.gvkey=b.gvkey and b.linkprim in ('P', 'C') and b.LINKTYPE in ('LU', 'LC') and (b.LINKDT <= a.fyenddt or missing(b.LINKDT)) and (a.fyenddt <= b.LINKENDDT or missing(b.LINKENDDT)) order by gvkey, datadate; quit; * Add CRSP monthly return data; proc sql; create table crsp1 as select a.*, b.date, b.ret label='' from combined_for_wrds as a, crsp.msf as b where a.permno = b.permno and intck('month', b.date, a.fybegdt) between 1 and 60 order by gvkey, datadate, date; quit; * Calculate monthly stock return volatility; proc means data=crsp1 noprint; var ret; by gvkey datadate year; output out=exec_roll_vol_fyear n=nrollingstd std=rollingstd; run; * Recreate riskfree_rate table; data rates; set frb.rates_daily; year=year(date); if &start_year <= year <= &end_year; keep date year TCMNOM_Y1 TCMNOM_Y2 TCMNOM_Y3 TCMNOM_Y5 TCMNOM_Y7 TCMNOM_Y10; run; proc means data=rates noprint; var TCMNOM_Y1 TCMNOM_Y2 TCMNOM_Y3 TCMNOM_Y5 TCMNOM_Y7 TCMNOM_Y10; by year; output out=rates_annual mean=oneyr twoyr threeyr fiveyr sevenyr tenyr; run; data riskfree_rate (drop=_TYPE_ _FREQ_); set rates_annual; oneyr=round(oneyr, 0.01); twoyr=round(twoyr, 0.01); threeyr=round(threeyr, 0.01); fiveyr=round(fiveyr, 0.01); sevenyr=round(sevenyr, 0.01); tenyr=round(tenyr, 0.01); fouryr=round(threeyr+(fiveyr-threeyr)/2, 0.01); sixyr=round(fiveyr+(sevenyr-fiveyr)/2, 0.01); eightyr=round(sevenyr+(tenyr-sevenyr)/3, 0.01); nineyr=round(sevenyr+(tenyr-sevenyr)/3*2, 0.01); run; * The following is Naveen's original program. I only change dataset references to make the program self-executable. * Please note I do not change Naveen's program comments (including dataset references in program comments). If you * want to run SAS codes in her program comments, you have to change them if appropriate. *==================================================================; * READING ANNCOMP DATASET - ANNUAL EXEC-LEVEL AGGREGATE COMPENSATION; *===================================================================; * this dataset has information on all execs, all years; * should have no duplicates by coperol-year; data anncomp; set execcomp.anncomp; * data was pulled as of Nov 2011 so we have only few observations for 2011 fiscal in execcomp; where &start_year <= year <= &end_year; coperol=co_per_rol; shrown=SHROWN_EXCL_OPTS; shrowntotal=SHROWN_TOT; opts_unvested_num=OPT_UNEX_UNEXER_NUM; opts_unvested_val=OPT_UNEX_UNEXER_EST_VAL; opts_vested_num=OPT_UNEX_EXER_NUM; opts_vested_val=OPT_UNEX_EXER_EST_VAL; opts_exercised_num= OPT_EXER_NUM; * see chk statements below - shrown_excl_opts is missing for some firms even though we have shrown_tot. But shrown_tot cannot be used even if shrown_excl_opts is missing because when both are there for a record, they are not very correlated; keep gvkey coperol year execid allothpd allothtot bonus eip_unearn_num eip_unearn_val noneq_incent old_datafmt_flag OPTION_AWARDS_BLK_VALUE OPTION_AWARDS_FV OPTION_AWARDS_NUM OPTION_AWARDS_RPT_VALUE OPTION_AWARDS opts_exercised_num opts_vested_val opts_vested_num opts_unvested_val opts_unvested_num OTHANN OTHCOMP RSTKGRNT SALARY SHROWN shrowntotal SHRS_VEST_NUM SHRS_VEST_VAL STOCK_AWARDS_FV STOCK_AWARDS STOCK_UNVEST_NUM STOCK_UNVEST_VAL TDC1 TITLEANN ltip ceoann defer_rpt_as_comp_tot; proc sort nodupkey; by coperol year; run; /** proc print data=anncomp (obs=100); var coperol year shrown tdc1; where coperol=2611; run; **/ /** coperol 16285 is palmisano of ibm - printing this out as a test case; proc print data=anncomp; where coperol=16285; var year old_datafmt_flag tdc1 OPTION_AWARDS_BLK_VALUE OPTION_AWARDS_FV OPTION_AWARDS_NUM opts_exercised_num opts_vested_val opts_vested_num opts_unvested_val opts_unvested_num shrown shrs_vest_num STOCK_AWARDS_FV STOCK_UNVEST_NUM STOCK_UNVEST_VAL ; title "summary stats for ibm palmisano"; run; **/ /** * tdc1old is the definition of how execcomp used to calculate tdc1; * post 2006, execcomp calculates tdc1 using tdc1new (although they continue to call it TDC1); * these are based on information frm S&P; * also, we verified that these give same values as tdc1; * OPTION_AWARDS and STOCK_AWARDS represent the expensed portion of total stock and option awards made to the CEO including stock and option awards from previous years (per FAS 123R); * we are dropping this going forward as they do not represnet the true annual compensation; * manual talks about two additional variables, OPT_UNEX_UNEXER and OPT_UNEX_EXER which represent the number of securities underlying the exerciseable and unexercisable options; * but these variables do not show up in the execcomp database; data chk1; set anncomp; tdc1old=salary+bonus+othann+rstkgrnt+ltip+allothtot +option_awards_blk_value; tdc1new=salary+bonus+noneq_incent+option_awards_fv + stock_awards_fv + defer_rpt_as_comp_tot + othcomp; difftdc1new=tdc1new-tdc1; difftdc1old=tdc1old-tdc1; run; * below, we find that mean, mix, max and median are all zero, which means we have the right formula for tdc1old and tdc1new above; proc means data=chk1 n mean median min max; var difftdc1old difftdc1new; title "difftdc1old and difftdc1new should be zero"; run; **/ /** * the chk below is to understand the old_datafmt_flag - this should be 0 post 2006; data chk2; set chk1; proc sort nodupkey; by gvkey year; proc sort ; by year; proc means noprint data=chk2; by year; var old_datafmt_flag ; output out=out1 mean= old_datafmt_flag ; run; proc print data=out1; title "all have oldformat=1 upto 2005, 15% have oldformat=1 in 2006, and all have oldformat=0 from 2007"; run; **/ *=================================================; * READING PERSON DATASET - EXEC-LEVEL INFORMATION; *=================================================; /** this dataset has information on executive name, gender, and age; * for our program here we only need the fullname so that we can check against proxy statements; * if necessary to confirm our understanding ; * should have no duplicates by execid;**/ data person; set execcomp.person; keep execid exec_fullname ; proc sort nodupkey; by execid; * proc contents ; * title "information on dataset person"; run; proc sql; create table anncomp1 as select * from anncomp x left join person y on x.execid=y.execid; proc sort nodupkey data=anncomp1; by coperol year; run; * ===========================; * READING IN CODIRFIN DATASET; * ===========================; * this dataset contains firm level information including aggregate director * information. So should not have duplicates by year; * data available from 1992-2010; * the bs_volatility and bs_yield were used by execcomp to calculate the BS values before 2007. After that they stopped calculating the BS values and started reporting the firm's estimates directly; * therefore they stopped reporting the bs_yield and the bs_volatility also after 2007; * we calculate the BS_values for the sake of consistency in the same manner before and after 2006. We therefore calculate the dividend yield and volatility using the same approach that execcomp adopted before 2007; * the dividend yield that goes into the BS value is the average of the last three years dividend yield and the volatility that goes into the BS value is based on estimates over 60 month rolling windows - see below; data codirfin; set execcomp.codirfin (keep = gvkey year bs_volatility bs_yield divyield shrsout prccf fyr srcdate); execcomp_volatility = bs_volatility; * given as % but it is used as fraction in formula; execcomp_yield=bs_yield/100; * found a mistake with dividend yield - for 2007, the dividend includes special dividend of $25 for Alberto Culver, GVKEY=001239. The annual report indicates that the dividend yield used for B-S (fair value) calculation is 1%. execcomp reports the dividend yield as 0.99%,0.96% and 101%. The last should actually be 0.66% ( =0.165/24.79). Using this, we get the dividend yield of 0.82% (average of last 3 years to be used in B-S, which gives us a more realistic calculated B_S value of $1,011 (in the ballpark of $920 reported by firm); if gvkey="001239" and year=2007 then divyield=0.66; * dividend yield should be a fraction for B_S calculation; divyield=divyield/100; keep gvkey year divyield execcomp_volatility execcomp_yield prccf shrsout srcdate fyr; proc sort nodupkey; by gvkey year; *proc contents data=common.codirfin; * title "information on dataset codirfin"; run; proc sort data=codirfin; by gvkey year; data codirfin; set codirfin; lgvkey=lag(gvkey); lyear=lag(year)+1; l2gvkey=lag2(gvkey); l2year=lag2(year)+2; ldivyield=lag(divyield); l2divyield=lag2(divyield); if gvkey^=lgvkey or year^=lyear then ldivyield=.; if gvkey^=l2gvkey or year^=l2year then l2divyield=.; if divyield^=. and ldivyield^=. and l2divyield^=. then estimated_yield= (divyield+ldivyield+l2divyield)/3; if divyield^=. and ldivyield^=. and l2divyield=. then estimated_yield= (divyield+ldivyield)/2; if divyield^=. and ldivyield=. and l2divyield=. then estimated_yield= divyield; if divyield=. and ldivyield=. and l2divyield=. then estimated_yield=.; drop ldivyield l2divyield lgvkey l2gvkey lyear l2year; run; /** proc print data=codirfin; where gvkey="006066"; title "print of codirfin for ibm"; run; **/ * ===========================; * READING IN COMBINED DATASET; * ===========================; * going to read in combined_201111_1_for_wrds. This was taken from combined crsp-compustat data from wrds and has fiscal year end date which we need later (in next step); * going to merge with codirfin because this dataset has only 1 obs per firm year; data combined; set combined_for_wrds(keep=gvkey permno year fybegdt fyenddt); where &start_year <= year <=&end_year; proc sort nodupkey; by gvkey year; run; /** proc means data=combined; title "combined crsp-compustat data from wrds"; run; **/ proc sql; create table codirfin1 as select * from codirfin x left join combined y on x.gvkey=y.gvkey and x.year=y.year; * should have no duplicates by gvkey-year; proc sort nodupkey data=codirfin1; by gvkey year; run; * ==========================; * SUBSET OF ANNCOMP DATASET ; * ==========================; /** do not need all of these variables beyond this point - we will only keep part of them and them when we are all done we can merge back the rest of the variables; This makes the program faster;**/ data anncomp2; set anncomp1; keep gvkey coperol year exec_fullname old_datafmt_flag option_awards_blk_value option_awards_num opts_exercised_num opts_vested_num opts_vested_val opts_unvested_num opts_unvested_val shrown; run; proc sql; create table excomp1 as select * from anncomp2 x left join codirfin1 y on x.gvkey=y.gvkey and x.year=y.year; data excomp2; set excomp1; * We want to make sure we pick up firms that have both execcomp and compustat data; * about 13,000 obs get dropped at this point; if fybegdt=. or fyenddt=. then delete; run; /** at this stage, we still have only 1 obs per executive per firm-year; * so this should have no duplicates by coperol-year;**/ proc sort nodupkey data=excomp2; by coperol year; run; * ===========================; * READING IN STGRTTAB DATASET; * ===========================; * this dataset has information on actual stock and option awards for the fiscal year; * this is available only from 1992-2006; * no duplicates by coperol grant number; data stgrttab; set execcomp.stgrttab; coperol=co_per_rol; keep coperol year blkshval exdate expric grntnum mktpric numsecur; proc sort nodupkey; by coperol year grntnum; *proc contents; * title "information on dataset stgrttab"; run; * at htis stage, for each executive per firm-year, we will have as many obs; * as the number of grants for that executive ; * so this can have more observations than excomp3; proc sql; create table excomp3 as select * from excomp2 x left join stgrttab y on x.coperol=y.coperol and x.year=y.year; * should have no duplicates by coperol-year-grntnum; proc sort nodupkey data=excomp3; by coperol year grntnum; run; * ===========================; * READING IN EX_BLACK DATASET; * ===========================; *execcomp used to provide the information on risk-free rate upto 2006 - they discontinuted after that; * this has data from 1992-2006 which is needed to do black-scholes values; * only 1 obs for each year - so totally 15 rows; * we are using this only to check our numbers with execcomp for the period when execcomp used to calculated BS values; * this is just to give confidence that the methodology we use to compute BS values is correct; * I am downloading the information from federal reserve website later for our calculation of BS values; data ex_black; set execcomp.ex_black; * risk free rate is given as a percentage, need it to be a fraction; execcomp_rf=risk_free_rate/100; keep execcomp_rf year; * proc contents; * title "information on dataset ex_black"; run; proc sql; create table excomp4 as select * from excomp3 x left join ex_black y on x.year=y.year; * =============================; * READING IN VOLATILITY DATASET; * =============================; * Here we calculate BS value of options granted using execcomp methodology and see if it matches their values reported; * execcomp uses divyield and volatility estimates over rolling 3 and 5 year periods in B-S calculations; * they stop providing this as of 2006, so we need to do it ourselves; * we do this in common.exec_roll_vol_fyear_201111_1.sas7bdat which uses crsp data to calculate volatilities of monthly stock returns; data volatility; set exec_roll_vol_fyear; where &start_year <= year <= &end_year; * we have volatility of monthly returns, converting to annual volatility; estimated_volatility=rollingstd*(12**0.5); proc sort nodupkey; by gvkey year; run; * ===========================================; * READING IN RISK-FREE RATES FROM FED WEBSITE; * ===========================================; * getting the risk-free rates based on Treasury securities; * source: fed reserve website (I downloaded to file in excel fed_website_10yrTnote); * I download the annual Treasury rates from 1992-2012. we use the risk free rate corresponding to the actual maturity; * note that execcomp assumes 7 year maturity of options and therefore use 7 year risk-free rate for grant date valuation; * note that FED website gives only 1,2,3,5,7 and 10 yr rates, I interpolate the numbers for 4,6,8,9 years; /* proc import OUT= riskfree_rate datafile= "c:/lalitha/execcomp/federal_reserve_treasury_rates.xlsx" DBMS=excel REPLACE; SHEET="federal_reserve_treasury_rates"; GETNAMES=YES; run; */ * proc print data=riskfree_rate; * run; * MERGE RISKFREE RATE DATA WITH VOLATILITY ESTIMATES; proc sql; create table BS_estimates as select * from volatility x left join riskfree_rate y on x.year=y.year; data BS_estimates1; set BS_estimates; * rates are in percentages, need to be in fractions; array X oneyr twoyr threeyr fouryr fiveyr sixyr sevenyr eightyr nineyr tenyr; do over X; X=X/100; end; keep estimated_volatility oneyr twoyr threeyr fouryr fiveyr sixyr sevenyr eightyr nineyr tenyr gvkey year nrollingstd; run; /** proc print data=BS_estimates1 ; where gvkey="001078"; title "check estimates of BS"; run; **/ proc sort nodupkey data=BS_estimates1; by gvkey year; run; proc sql; create table excomp5 as select * from excomp4 x left join BS_estimates1 y on x.gvkey=y.gvkey and x.year=y.year; *WINSORIZE DATA AS IN execcomp METHODOLOGY; *==========================================; * execcomp winsorizes values of volatility and dividend yield at the 5th and 95th percentile values when calculating B-S values, we follow same procedure; proc sort data=excomp5; by year; proc univariate data=excomp5 noprint; by year; var estimated_yield estimated_volatility; output out=out1 pctlpts=5 95 pctlpre=estimated_yield estimated_volatility; data out1; set out1; keep year estimated_yield5 estimated_yield95 estimated_volatility5 estimated_volatility95; run; proc sort data=excomp5; by year; proc sort data=out1; by year; data excomp6a; merge excomp5 (in=A) out1 (in=B); by year; if A; * proc print data=excomp6 (obs=10); * var gvkey year stimated_volatility5 estimated_volatility95 estimated_yield5 estimated_yield95; * title "check winsorizing done"; run; data excomp6b; set excomp6a; if estimated_yield>estimated_yield95 and estimated_yield^=. then estimated_yield=estimated_yield95; if estimated_volatility>estimated_volatility95 and estimated_volatility^=. then estimated_volatility=estimated_volatility95; if estimated_yield<estimated_yield5 and estimated_yield^=. then estimated_yield=estimated_yield5; if estimated_volatility<estimated_volatility5 and estimated_volatility^=. then estimated_volatility=estimated_volatility5; drop estimated_volatility5 estimated_volatility95 estimated_yield5 estimated_yield95; run; * computing annual volatililities for using when firms have less than 1 year of data; * using the mean estimated volatility for that sample year as in execcomp methodology; proc sql; create table excomp6c as select *, mean(estimated_volatility) as mean_vol, mean(estimated_yield) as mean_yield from excomp6b group by year; /* * the datastep below shows that our volatility numbers are slightly higher than the sp1500 numbers for the period 1992-2003 for which they show the data on their website; * our range is from 0.34-0.53 and their range is from 0.31-0.50; * they do not show the average div yield in their sample, only the 95th percentile values; data chk3; set excomp6c; proc sort nodupkey; by year; proc print data=chk3; var year mean_vol mean_yield; title "estimated annual volatilities and yields for sp1500 firms"; run; */ data excomp7; set excomp6c; format exdate mmddyy10.; * if we have less than 1 year of data, we use the average volatility of S&P1500 firms following execcomp methodology; * for this, I need the averages by year - execcomp reports the numbers they use upto 2003. we use the numbers based on our own estimates for all years (see module above); if nrollingstd<12 and nrollingstd^=. then estimated_volatility=mean_vol; * same for estimated_volatility - use execcomp volatility upto 2006 if ours is missing, after which they do not report; if year<=2006 and estimated_volatility=. then estimated_volatility=execcomp_volatility; * since we are using the execcomp dividend yield numbers to calculate the 3-year rolling averages, we can actually do the computation only for the 3rd year that each firm is there on execcomp; * to avoid missing values, we use the actual execcomp 3-year averages till 2006, and use the estimated ones from 2007 onwards; if year<=2006 and estimated_yield=. then estimated_yield=execcomp_yield; /* * execcomp assumes that all grants are issued on july 1st of grant year; fiscal data assumed grant date year year for B-S method end ===== ==== ====== Jan-06 2005 Jul-05 Feb-06 2005 Jul-05 Mar-06 2005 Jul-05 Apr-06 2005 Jul-05 May-06 2005 Jul-05 Jun-06 2006 Jul-05 Jul-06 2006 Jul-06 Aug-06 2006 Jul-06 Sep-06 2006 Jul-06 Oct-06 2006 Jul-06 Nov-06 2006 Jul-06 Dec-06 2006 Jul-06 */ if fyr=6 then assumed_grantyear=year-1; else assumed_grantyear=year; assumed_grantdate=mdy(7,1,assumed_grantyear); * need to get fiscal year end dates for maturity at fiscal year end; * two maturity values need to be estimated; * one for B-S value that goes in the TDC1 (as of grant date) and the other for B-S value that goes into to delta (as of fiscal year end); * this is based on difference between exdate and grantdate and between exdate and fiscal year end date respectively; * execcomp rounds off maturity to nearest whole number; * WE ROUND OFF IN DATA CHECKS ONLY TO MAKE SURE WE GET OVER A 0.999 CORRELATION WITH execcomp VALUES = GOING FORWARD WE DO NOT ROUND OFF ANYWHERE; * ALSO IMPORTANT - EXECUOMP has a 70% HAIRCUT ON TIME TO MATURITY; * THIS IS BECAUSE EXECS TYPICALLY EXERCISE EARLIER; * CORE AND GUAY DO NOT APPPEAR TO MAKE THIS ASSUMPTION; * IF WE WISH TO DO A 70% HAIRCUT, DO THE STATEMENTS BELOW; * maturity_grantdate=0.7*round((exdate-assumed_grantdate)/365,1); * maturity_yearend=0.7*round((exdate-fyenddt)/365,1); * IF WE WISH TO USE ACTUAL MATURITY RATHER THAN 70% HAIRCUT, DO THE STATEMENTS BELOW; maturity_grantdate=(exdate-assumed_grantdate)/365; maturity_yearend=(exdate-fyenddt)/365; if coperol=. then delete; * see note below for dataset chk4; * zero observations where maturity at grant date is negative; * but 132 cases where maturity at yearend is negative; * seems reasonable to assume maturity is very small - cannot take zero because this comes in denominator of delta - assume very small maturity; if numsecur^=. and numsecur>0 and maturity_yearend<0 and maturity_yearend^=. then maturity_yearend=0.001; drop assumed_grantyear assumed_grantdate; run; /* * if the maturity is negative (happens when we do fiscal year end), it means that the expiry date is before fiscal year end; * I find 0 obs that have maturity_grantdate<0; * there are 132 obs that have maturity_yearend<0; * appears to be cases where options were given to vest immediately; * checked two of them in proxy statements (see aaron rents in 2005, exec todd evans, coperol 25896 and J Bryant, Kellogg, coperol 28564; * so these are coded as maturity=. since we do not know if they * have been exercised and converted to shares (we will then be doublecounting delta); * obviously this check has to be done before the final step in the previous program is done, which replaces the negatives with zeros; data chk4; set excomp7; proc sort ; by coperol descending year; proc print ; var coperol year fyr fyenddt exdate maturity_grantdate maturity_yearend; * where numsecur^=. and numsecur>0 and maturity_grantdate<0 and maturity_grantdate^=.; where numsecur^=. and numsecur>0 and maturity_yearend<0 and maturity_yearend^=.; format fyenddt mmddyy10.; title "maturity is negative"; run; proc means data=excomp7; var exdate fyenddt maturity_grantdate maturity_yearend ; where numsecur>0 and numsecur^=.; title "out of 127861 obs for which grants were made, 127353 had exdate available"; run; proc sql; create table chk5 as select *, count(coperol) as ctcoperol, count(maturity_grantdate) as ctmat_grantdate from excomp7 group by coperol, year; proc means data=chk5; var maturity_grantdate; title "checking before we replace missing maturity values"; run; */ *sometimes when there are multiple grants for the same executive in the same year, data on maturity of one option award is missing even though the other awards in that year have data on maturity; * in these cases I am going to replace the maturity that is missing with the average maturity of the other awards granted to the same executive in the same year; * number of obs goes from 127355 to 127523 as per check statement above and below; * this changes the mean from 9.0556 to 9.0559; proc sql; create table excomp8 as select *, mean(maturity_grantdate) as avg_mat_grantdate, mean(maturity_yearend) as avg_mat_yearend from excomp7 group by coperol, year; data excomp9; set excomp8; if maturity_grantdate=. and avg_mat_grantdate^=. then maturity_grantdate=avg_mat_grantdate; if maturity_yearend=. and avg_mat_yearend^=. then maturity_yearend=avg_mat_yearend; run; /** proc means data=excomp9 n mean min p1 p5 p10 p25 p50 p75 p90 p99 max; var maturity_grantdate maturity_yearend; title "checking after we replace missing maturity values"; run; **/ * we can use either the execcomp values or the estimated (by us) values of risk-free rate, volatility and divyield; * remember that the risk free rate depends on the maturity of the option at year end or grant date; data excomp10; set excomp9; * USE THESE NUMBERS IF YOU WANT TO USE PRE-2006 execcomp VALUES; * I USE THIS ONLY TO CORRELATE OUR CALCULATED VALUES WITH execcomp BS VALUES; * THIS WILL BE VALID ONLY TILL 2006; * rfc=execcomp_rf; * bs_yield=execcomp_yield; * sigma=execcomp_volatility; if round(maturity_yearend,1) =1 then rfc=oneyr; if round(maturity_yearend,1) =2 then rfc=twoyr; if round(maturity_yearend,1) =3 then rfc=threeyr; if round(maturity_yearend,1) =4 then rfc=fouryr; if round(maturity_yearend,1) =5 then rfc=fiveyr; if round(maturity_yearend,1) =6 then rfc=sixyr; if round(maturity_yearend,1) =7 then rfc=sevenyr; if round(maturity_yearend,1) =8 then rfc=eightyr; if round(maturity_yearend,1) =9 then rfc=nineyr; if round(maturity_yearend,1)=10 then rfc=tenyr; * sometimes maturity is greater than 10; if (maturity_yearend^=. and round(maturity_yearend,1)>10) then rfc=tenyr; bs_yield=estimated_yield; sigma=estimated_volatility; sigmasq=sigma*sigma; drop execcomp_rf execcomp_yield execcomp_volatility estimated_yield estimated_volatility; run; /** proc print data=excomp10; var year rfc sigma sigmasq maturity_yearend maturity_grantdate; where coperol=16285; title "ibm checking recall no data will be there post-2006 on options"; run; **/ *=========================================; * CALCULATE INCENTIVES FOR PRE-2006 PERIOD; *=========================================; * START WITH SENSITIVITY OF CURRENT YEAR OPTIONS; *------------------------------------------------; * FOR THIS YEAR'S GRANT we can use blk_val as the value of grants this year or calulate as per core-guay; * execcomp uses 'mktprice' which is the market price on the date of issue of the option to calculate B-S value; * i do the same for the grant date, but use market price on fiscal year close (prccf) as per Core and Guay for the fiscal year end price; * execcomp assumes that maturity is only 0.7* maturity as given by exdate; * reason - these options are usually exercised earlier; * as mentioned above, WE DO NOT use same assumption, but this can be changed easily in the program; * I found the value as per the formula below for Vopts_grantdate, using the execcomp assumptions and compared it to blk_val, find that the two numbers are very close; data excomp11; set excomp10; * sometimes, it appears that the numsecur is missing when it is actually zero; * if we compare our nubmers with the numbers in execcomp, they get more observations of option_awards_num compared to our sumnumsecur; * to correct for this, i am going to code numsecur=0 if numsecur=. and if the other data on option awards is missing; if numsecur=. and (exdate=. and expric=. and mktpric=. and option_awards_num=0) then numsecur=0; * has to be done before we throw out the duplicate coperols; * exercise price; Xc = expric; * REALIZABLE VALUE AS EXCESS OF S OVER X; realizable_value = (prccf-Xc)*numsecur; if realizable_value<0 then realizable_value=0; if mktpric=. or Xc=. or numsecur=. then realizable_value=.; * we will calculate two values of Zc, one for B-S value at grant date. This is for TDC1. The other is for B-S value at year end. This is for delta and vega; Zc_yearend = (log(prccf/Xc)+maturity_yearend*(rfc-bs_yield+sigmasq/2))/(sigma*sqrt(maturity_yearend)); if maturity_yearend=. then Zc_yearend=.; Zc_grantdate= (log(mktpric/Xc)+maturity_grantdate*(rfc-bs_yield+sigmasq/2))/(sigma*sqrt(maturity_grantdate)); if maturity_grantdate=. then Zc_grantdate=.; * appendix of core guay paper; Sc_yearend = exp(-bs_yield*maturity_yearend)*probnorm(Zc_yearend)*numsecur*prccf/100; if Zc_yearend=. then Sc_yearend=.; * If it is missing, then we are assuming no option grants were made, and hence zero; * if numsecur=. or numsecur=0 then Sc_yearend=0; if numsecur=0 then Sc_yearend=0; * Black Scholes value of options; Vc_grantdate = numsecur*(mktpric*exp(-bs_yield* maturity_grantdate)*probnorm(Zc_grantdate) -Xc*exp(-rfc* maturity_grantdate)*probnorm(Zc_grantdate-sigma*sqrt( maturity_grantdate))); if Zc_grantdate=. then Vc_grantdate=.; * If it is missing, then we are assuming no option grants were made, and hence zero; * if numsecur=. or numsecur=0 then Vc_grantdate=0; if numsecur=0 then Vc_grantdate=0; * introduced for our pay for luck paper, to find the wealth measures; if Zc_yearend^=. then do; Vc_yearend = numsecur*(prccf*exp(-bs_yield*maturity_yearend)*probnorm(Zc_yearend) -Xc*exp(-rfc*maturity_yearend)*probnorm(Zc_yearend-sigma*sqrt(maturity_yearend))); end; * If it is missing, then we are assuming no option grants were made, and hence zero; * if numsecur=. or numsecur=0 then Vc_yearend=0; if numsecur=0 then Vc_yearend=0; * sensitivity with respect to a 0.01 change in stock return volatility; * see core and guay appendix A; if Zc_yearend^=. then do; Rc_yearend = exp(-bs_yield* maturity_yearend)*PDF('normal',Zc_yearend,0,1)*prccf*sqrt( maturity_yearend)*0.01*numsecur; end; if Zc_yearend=. or maturity_yearend=. then Rc_yearend=.; * If it is missing, then we are assuming no option grants were made, and hence zero; * if numsecur=. or numsecur=0 then Rc_yearend=0; if numsecur=0 then Rc_yearend=0; run; /** * note that proc print below is correct only upto 2005; * after 2005, ibm is on new reporting format, so reports 2006 onward data in different dataset; proc print data=excomp11; * where coperol=16285; where coperol=6; var year rfc bs_yield sigma numsecur maturity_yearend Xc Zc_yearend Sc_yearend Rc_yearend Vc_yearend old_datafmt_flag; title "ibm palmisano check of current year option calculation"; run; **/ * AGGREGATING TRANCHE-LEVEL INFORMATION TO PROVIDE 1 OBS PER EXECUTIVE; * since each executive can be given more than 1 option grant in 1 year, I need to cumulate the value calculated above for each coperol year combination; proc sql; create table excomp12 as select *, sum(Vc_grantdate) as Vopts_grantdate, sum(Vc_yearend) as Vopts_yearend, sum(numsecur) as sumnumsecur, sum(Sc_yearend) as Sopts_grants_yearend, sum(Rc_yearend) as Ropts_grants_yearend, sum(realizable_value) as sumrealizable_value from excomp11 group by coperol,year; /* * when we do this with execcomp 70% haircut on grant date maturity as well as their measures of rf and dividend yield and volatility the difference between the two values has a mean of 2%, median of 0%, 10th and 90th of 0% and 5.5% and the correlations between our estimated values and execcomp estimated values are about 0.998; * no difference in option number total, and the correlation between our sumnumsecur and option_awards_num=1; * when we do this with execcomp 70% cut and with our measures of rf and dividend yield and volatility the difference between the two values has a mean of 11%, median of 7.5%, and 10th and 90th percentiles of 1.3% and 22% and the correlations between our estimated values and execcomp estimated values is 0.992 ; data chk6; set excomp12; diff_optvalue=abs(option_awards_blk_value-Vopts_grantdate)/option_awards_blk_value; diff_optnum=abs(option_awards_num-sumnumsecur)/option_awards_num; keep coperol year diff_optvalue exdate maturity_grantdate maturity_yearend Vc_grantdate rfc Vopts_grantdate sumnumsecur option_awards_blk_value option_awards_num blkshval diff_optnum; proc sort nodupkey; by coperol year; run; proc univariate data=chk6; var diff_optvalue diff_optnum; where year<=2005; title "comparing our values with execcomp"; run; proc corr data=chk6; var option_awards_blk_value Vopts_grantdate; title "correlation between our calculated value and execcomp B-S value is over 99.8% using execcomp numbers"; where year<=2005; run; proc corr data=chk6; var sumnumsecur option_awards_num; title "correlation between our calculated value and reported value in execcomp of total number of options is 100% but number of obs different"; where year<=2005; run; * number of observations between sumnumsecur (our calculated total annual grants and option_awards_num (execcomp provided number of total grants) are different for 21653 cases; * these are cases for which execcomp does not calculate the BS value anyway, i.e., the option_awards_blk_value is missing, so it appears that these are problematic cases anyway; * so should be ok to ignore them; proc print data=chk6 (obs=25); where sumnumsecur=. and year<=2005; run; proc univariate data=chk6; var option_awards_num option_awards_blk_value vopts_grantdate; where sumnumsecur=. and year<=2005; title "option_awards_blk_value is missing when sumnumsecur is missing but option_awards_num is not missing"; title "there are about 22151 obs that have this problem"; title " we can treat these as missing or zero, we leave it as missing"; run; */ * at this point, we are going to keep only 1 observation per executive year; * since we have already added up the values of current year's grants, that is all we need to keep; * going forward, it should always be 1 obs per executive year; data excomp13; set excomp12; if sumnumsecur=0 or sumnumsecur=. then do; Sopts_grants_yearend=0; Vopts_grantdate=0; Ropts_grants_yearend=0; sumrealizable_value=0; Vopts_yearend=0; FVopts_yearend=0; end; * dropping rfc here, recalculating rf below based on matvest and matunvest; * the only reason to calculate Vopts_grantdate was to compare with execcomp; * so not keeping this below, not keeping BS values reported by execcomp either; keep coperol year Vopts_yearend sumnumsecur Sopts_grants_yearend Ropts_grants_yearend sumrealizable_value sigma sigmasq bs_yield option_awards_num opts_vested_num opts_unvested_num opts_exercised_num opts_vested_val opts_unvested_val shrown shrsout old_datafmt_flag avg_mat_yearend prccf fybegdt fyenddt oneyr twoyr threeyr fouryr fiveyr sixyr sevenyr eightyr nineyr tenyr exec_fullname; proc sort nodupkey; by coperol year; run; * first take lags of number of vested and unvested options; * we use these below to check, see comments; proc sort data=excomp13; by coperol year; data excomp13a; set excomp13; lcoperol=lag(coperol); lyear=lag(year)+1; lag_opts_vested_num=lag(opts_vested_num); lag_opts_unvested_num=lag(opts_unvested_num); if coperol^=lcoperol or year^=lyear then do; lag_opts_vested_num=. ; lag_opts_unvested_num=.; end; drop lcoperol lyear ; run; /* * compare total number of options and total value before and after this exercise; * in general, the number of options at the start of the year (lagged values * of opts_unvested_num and VESTED NUM) plus additional options granted less * options exeercised should be equal to the number of options at the * end of the year (sum of opts_vested_num and opts_unvested_num); * i.e., lag(opts_unvested_num)+lag(opts_vested_num)+OPTION_AWARDS_NUM = opts_exercised_num + opts_unvested_num + opts_vested_num; * appears that this is not always true because of stock splits; * specifically, in 18% of cases, this is not true; * splits will not affect us as the sensitivties etc depend on both price * and number of options, both of which are adjusted; * but as in the case of AMR, ALCOA etc, this may mean that we do not find * the above equality to hold; * also may be the case when there are expired options - not sure how to capture that; * also check to see if any of the vested or unvested num or value is negative; * should never be, but I find about 1 to 17 obs have this problem; * set them to zero in the next step; data chk7; set excomp13a; balance = lag_opts_unvested_num + lag_opts_vested_num + option_awards_num - (opts_exercised_num + opts_unvested_num + opts_vested_num); * flag if the balance is greater than 10. It should be zero, doing 0.010 or 10 options to allow for rounding off issues; if abs(balance)>0.010 then flag=1; else flag=0; if opts_unvested_num<0 and opts_unvested_num^=. then flag1=1 ; else flag1=0; if opts_unvested_val<0 and opts_unvested_val^=. then flag2=1 ; else flag2=0; if opts_vested_num<0 and opts_vested_num^=. then flag3=1 ; else flag3=0; if opts_vested_val<0 and opts_vested_val^=. then flag4=1 ; else flag4=0; run; proc freq data=chk7; tables flag flag1 flag2 flag3 flag4; where year<=2005; title "flag should be zero typically and flag1-flag4 should be 0"; run; proc means data=chk7 n mean p5 p10 p25 p50 p75 p90 p95; var balance; title "balance should be zero, but is not. this is because of splits, expiry etc - see note above"; run; */ *=======================================================; * NOW COMPUTE SENSITIVITY OF UNVESTED AND VESTED OPTIONS; *=======================================================; * FOR STOCK OF OPTION GRANTS, DO THIS; * first, we need the average number of exercisable options, based on the total realizable value of exercisable options reported in the proxy; * the realizable value is the excess of stock over exercise price for all exercisable options; * then we subtract this from the stock price, which would give the average exercise price; * market price used is the average year end market price; * set time to maturity for unexercisable options as one year less than the time to maturity for current year options (or 9 if no options in current year); * do the same thing for exercisable options, but keep the time to maturity as three years less than the time to maturity for unexercisable options; * subtract current years options from unexercised options portfolio to aviod double counting; data excomp13b; set excomp13a; * see chk step above to see if any of the vested or unvested num or value is negative. It should never be, but I find about 10 obs have this problem. I set them to zero below; if opts_unvested_num<0 and opts_unvested_num^=. then opts_unvested_num=0; if opts_unvested_val<0 and opts_unvested_val^=. then opts_unvested_val=0; if opts_vested_num<0 and opts_vested_num^=. then opts_vested_num=0; if opts_vested_val<0 and opts_vested_val^=. then opts_vested_val=0; *===============; * unexercisable; *===============; * * also, if we assume stock options granted in the year are all unexercisable; * then opt_awards_num should be strictly smaller than OPT_UNEX_UNEXER_NUM; * not always true however (not true in xxx cases); * looks like some options immediately vest; * therefore we cannot assume current years options have long vesting; * if options issued in current year, then time to maturity is maturity-1, else 9; * also, the unexercisable option portfolio has to be reduced by the number of options granted that year and the value of options granted that year should be reduced from total realizable value of unexercisable options ; * need to adjust the unexercisable options by the number granted that year; * that's because we separately calculate the deltas for these; * assuming all current year's options are unvested; opts_unvested_num_excl_curryear = opts_unvested_num - option_awards_num; * sumrealizable_value represents the aggregate value of all the options awarded that year to each exec; * so it is numsecur*(fiscal year end price - exercise price) aggregated across all stock grants during year; * by subtracting this from opts_unvested_val, we get the aggregate value of unvested options that were granted in prior years (which = opts_unvested_val_excl_curryear); * for example, assume that UNVEST_NUM=40, OPTION_AWARDS_NUM=50; * that means that some of the awards given this year vested immediately; * first we assume that these were not exercised, then we adjust for possible exercise; * see notes below; * in the example here, we reduce the current year's portfolio of unvested options to zero, because we are separetly considering the 50 options that were granted; * the 10 options that were given in the current year are reduced from the stock * of vested options; opts_unvested_val_excl_curryear = opts_unvested_val-sumrealizable_value; opts_vested_num_excl_curryear = opts_vested_num; opts_vested_val_excl_curryear = opts_vested_val; if opts_unvested_num_excl_curryear^=. and opts_unvested_num_excl_curryear<0 then do; opts_unvested_num_excl_curryear=0; opts_unvested_val_excl_curryear =0; opts_vested_num_excl_curryear= opts_vested_num + (opts_unvested_num - option_awards_num); opts_vested_val_excl_curryear= opts_vested_val + (opts_unvested_val-sumrealizable_value); end; if opts_unvested_val_excl_curryear^=. and opts_unvested_val_excl_curryear<0 then opts_unvested_val_excl_curryear=0; run; /* proc means data=excomp13b n mean p5 p25 p50 p75 p95; var opts_unvested_num_excl_curryear; where opts_unvested_num_excl_curryear<0 and opts_unvested_num_excl_curryear^=.; title "should be no observations here"; run; proc means data=excomp13b n mean p5 p25 p50 p75 p95; var opts_unvested_val_excl_curryear; where opts_unvested_val_excl_curryear<0 and opts_unvested_val_excl_curryear^=.; title "should be no observations here"; run; proc means data=excomp13b n mean p5 p25 p50 p75 p95; var opts_vested_num_excl_curryear; where opts_vested_num_excl_curryear<0 and opts_vested_num_excl_curryear^=.; title "should be no observations here"; run; proc means data=excomp13b n mean p5 p25 p50 p75 p95; var opts_vested_val_excl_curryear; where opts_vested_val_excl_curryear<0 and opts_vested_val_excl_curryear^=.; title "should be no observations here"; run; */ * checked that all opts_unvested_num_excl_curryear and opts_unvested_val_excl_curryear are now zero or positive; * checked that all opts_unvested_num are now zero or positive; * some cases (3790 obs) of opts_vested_num are now negative; * these have to be cases where all or part of current years * grant was exercisable immediately, and some were actually exercised; * in this case, we check if overall equality (see earlier comment) holds; * if not, drop these obs becuase it is not clear what is going on (even from proxy); * ROUND function below matches at 100th place-that is if the difference is less than 100 options, we consider it ok; * this is only to make sure that rounding off error is not responsible for failure to match; * if the equality holds, then we need to replace opts_vested_num_excl_curryear to zero; data excomp13c; set excomp13b; if opts_vested_num_excl_curryear<0 and opts_vested_num_excl_curryear^=. then if round((opts_unvested_num + opts_vested_num),0.1)=round((LAG_opts_unvested_num + LAG_opts_vested_num + OPTION_AWARDS_NUM - opts_exercised_num),0.1) then do; opts_vested_num_excl_curryear=0; opts_vested_val_excl_curryear=0; DUMMY1=1; end; * if the equality does not hold, then we need to replace all values to missing; if opts_vested_num_excl_curryear<0 and opts_vested_num_excl_curryear^=. then if round((opts_unvested_num + opts_vested_num),0.1)^=round((LAG_opts_unvested_num + LAG_opts_vested_num + OPTION_AWARDS_NUM - opts_exercised_num),0.1) then do; opts_unvested_num_excl_curryear=.; opts_unvested_val_excl_curryear=.; opts_vested_num_excl_curryear=.; opts_vested_val_excl_curryear=.; DUMMY2=1; end; * after these adjustments, still opts_vested_val_excl_curryear is negative for some cases, although for some it is just rounding off error, going to make this zero for everyone; * may be because the dates/assumptions we use for SUMREALIZABLE_VALUE are slightly * different from dates used by company; if opts_vested_val_excl_curryear<0 and opts_vested_val_excl_curryear^=. then opts_vested_val_excl_curryear=0; run; /** proc means data=excomp13c ; var dummy1 dummy2; title "checking how many observations had vested_num or vested_val is negative"; run; **/ data excomp14; set excomp13c; * maturity of unercisable set at 9 years or actual minus 1; if option_awards_num^=. and option_awards_num>0 then matunvest=avg_mat_yearend-1; else matunvest=9; if matunvest<0 and matunvest^=. then matunvest=0.001; * dont make it zero because the Z function uses this in denominator; * if matunvest<0 then matunvest=1; if round(matunvest,1) <=1 then rfunvest=oneyr; if round(matunvest,1) =2 then rfunvest=twoyr; if round(matunvest,1) =3 then rfunvest=threeyr; if round(matunvest,1) =4 then rfunvest=fouryr; if round(matunvest,1) =5 then rfunvest=fiveyr; if round(matunvest,1) =6 then rfunvest=sixyr; if round(matunvest,1) =7 then rfunvest=sevenyr; if round(matunvest,1) =8 then rfunvest=eightyr; if round(matunvest,1) =9 then rfunvest=nineyr; if round(matunvest,1)=10 then rfunvest=tenyr; * sometimes maturity is greater than 10; if (matunvest^=. and round(matunvest,1)>1) then rfunvest=tenyr; * if Xunvest<0 then presumably the option is deep in the money; * so we reset the exercise price to be very small; if opts_unvested_num_excl_curryear^=0 then Xunvest=prccf-(opts_unvested_val_excl_curryear/opts_unvested_num_excl_curryear); if (Xunvest<0 and Xunvest^=.) then Xunvest=0.01; Zunvest = (log(prccf/Xunvest)+ matunvest*(rfunvest-bs_yield+sigmasq/2))/(sigma*sqrt(matunvest)); if matunvest=. or matunvest<0 then Zunvest=.; if opts_unvested_val_excl_curryear=. or opts_unvested_val_excl_curryear<0 or opts_unvested_num_excl_curryear<0 or opts_unvested_num_excl_curryear=. then Zunvest=.; if opts_unvested_val_excl_curryear=. or opts_unvested_val_excl_curryear<0 or opts_unvested_num_excl_curryear<0 or opts_unvested_num_excl_curryear=. then FZunvest=.; Sunvest = exp(-bs_yield* matunvest)*probnorm(Zunvest)*(prccf/100)*opts_unvested_num_excl_curryear; * sensitivity with respect to a 0.01 change in stock return volatility; * see core and guay appendix A; if Zunvest^=. then do; Runvest = exp(-bs_yield* matunvest)*PDF('normal',Zunvest,0,1)*prccf*sqrt( matunvest)*0.01*opts_unvested_num_excl_curryear; end; if Zunvest=. then Runvest=.; *value of option portfolio ($000); Vunvest = opts_unvested_num_excl_curryear*(prccf*exp(-bs_yield* matunvest)*probnorm(Zunvest) -Xunvest*exp(-rfunvest* matunvest)*probnorm(Zunvest-sigma*sqrt( matunvest))); if opts_unvested_num_excl_curryear=. or Zunvest=. then Vunvest=.; *============; * exercisable; *============; * maturity is 3 years less than that of unexercisable - dont make it zero because the Z function uses this in denominator; matvest=matunvest-3; if matvest<0 and matvest^=. then matvest=0.001; Xvest=prccf-(opts_vested_val_excl_curryear/opts_vested_num_excl_curryear); * if Xvest<0 then presumably the option is deep in the money; * so we reset the exercise price to be very small; if (Xvest<0 and Xvest^=.) then Xvest=0.01; if round(matvest,1) <=1 then rfvest=oneyr; if round(matvest,1) =2 then rfvest=twoyr; if round(matvest,1) =3 then rfvest=threeyr; if round(matvest,1) =4 then rfvest=fouryr; if round(matvest,1) =5 then rfvest=fiveyr; if round(matvest,1) =6 then rfvest=sixyr; if round(matvest,1) =7 then rfvest=sevenyr; if round(matvest,1) =8 then rfvest=eightyr; if round(matvest,1) =9 then rfvest=nineyr; if round(matvest,1)=10 then rfvest=tenyr; * sometimes maturity is greater than 10; if (matvest^=. and round(matvest,10)>10) then rfvest=tenyr; Zvest = (log(prccf/Xvest)+ matvest*(rfvest-bs_yield+sigmasq/2))/(sigma*sqrt(matvest)); if matvest=. or matvest<0 then Zvest=.; if opts_vested_val_excl_curryear=. or opts_vested_val_excl_curryear<0 or opts_vested_num_excl_curryear<0 or opts_vested_num_excl_curryear=. then Zvest=.; * sensitivity of option to 1 percent change in stock price; * appendix of core guay paper; Svest = exp(-bs_yield* matvest)*probnorm(Zvest)*(prccf/100)*opts_vested_num_excl_curryear; * sensitivity to stock return volatility; if Zvest^=. then do; Rvest = exp(-bs_yield* matvest)*PDF('normal',Zvest,0,1)*prccf*sqrt( matvest)*0.01*opts_vested_num_excl_curryear; end; if Zvest=. then Rvest=.; Vvest = opts_vested_num_excl_curryear*(prccf*exp(-bs_yield*matvest)*probnorm(Zvest) -Xvest*exp(-rfvest*matvest)*probnorm(Zvest-sigma*sqrt(matvest))); if opts_vested_num_excl_curryear=. or Zvest=. then Vvest=.; * sensitivity of shareholdings; Sshr= shrown*prccf/100; * remember that shrown is also in 000s; Vshr = shrown*prccf; * sometimes both opts_unvested_val_excl_curryear and opts_unvested_num_excl_curryear are zero; * in this case, the Xunvest cannot be computed because opts_unvested_val_excl_curryear/opts_unvested_num_excl_curryear is not defined; * for example, for ABBOTT LABS in 1998, the # current year stock option grants was 1944.36, but # unvested options at end of year was only 1399.22, so we adjust the number of unvested excluding current year to be zero, then the number of vested is adjusted downward from the reported value of 1100 to (1100-(1944-1399)) = 555; * in this case, we still get Vunvest, Runvest etc. to be missing, so we have to set it equal to zero; * overall sensitivity; * we do not want to make delta=. if one of the components is missing; if opts_vested_num_excl_curryear=0 then do; Svest=0; Rvest=0; Vvest=0; end; if opts_unvested_num_excl_curryear=0 then do; Sunvest=0; Runvest=0; Vunvest=0; end; *execcomp reports restricted stock where possible along with the sharebased holdings; * so we do not include restricted stock as part of the delta; * email from execcomp: "For (stock ownership) item we are using the Stock-Based Holdings column which includes the restricted stock. We have also made corrections to past years. As with much of the execcomp database, what we collect is largely dependent on how the company reports the data. When possible we will try to include restricted stock with this item, however sometimes the company doesn't provide a concise picture and it's not possible"; delta = sum(Svest, Sunvest, Sopts_grants_yearend, Sshr); optiondelta = sum(Svest, Sunvest, Sopts_grants_yearend); sharedelta = Sshr; Ropt = sum(Rvest, Runvest, Ropts_grants_yearend); * firm-related wealth; Vportfolio_yearend = sum(Vvest, Vunvest, Vopts_yearend, Vshr); *IMPORTANT = WE TAKE VARIABLES SUCH AS NUMBER OF VESTED AND UNVESTED FROM ANNCOMP TABLE AS SUCH THESE ARE AVAILABLE FOR BOTH OLD AND NEW FORMAT DATA BECAUSE ANNCOMP PRESENTS AGGREGATES. SO FOR EXAMPLE, WE USE THE DATA VARIABLE OPTS_VESTED_NUM AND OPTS_UNVESTED_NUM FOR THE PORTFOLIO OF VESTED AND UNVESTED USING OLD METHOD. BUT THIS IS AVAILABLE POST-2006. ALSO, WE ASSUME MATUNVEST=9 IF NO OPTION GRANTS IN CURRENT YEAR. GIVEN THIS, AND OTHER ASSUMPTIONS WE MAKE, WE COULD END UP CALCULATING VEGA AND DELTA FOR THE OBSERVATIONS WHICH HAVE DETAILED DATA IN OUTSTANDINGAWARDS TABLE. TO MAKE SURE WE DO NOT DO THAT, USE THE FOLLOWING STEP; if old_datafmt_flag=0 then do; Svest=.; Sunvest=.; Sopts_grants_yearend=.; Sshr=.; Ropt=.; Rvest=.; Runvest=.; Ropts_grants_yearend=.; Vportfolio_yearend=.; Vvest=.; Vunvest=.; Vopts_yearend=.; Vshr=.; delta=.; end; run; /** proc print data=excomp14; where coperol=16285; * var Zvest prccf Xvest matvest rfvest bs_yield sigmasq sigma matvest; var year bs_yield matunvest Xunvest matvest Xvest Zvest sumnumsecur opts_unvested_num opts_unvested_val opts_unvested_num_excl_curryear opts_unvested_val_excl_curryear opts_vested_num_excl_curryear opts_vested_val_excl_curryear Vopts_yearend Svest Sunvest Sopts_grants_yearend sshr Rvest Runvest Ropts_grants_yearend ropt delta ; title "ibm palmisano check of current year option calculation"; run; **/ /* proc print data=excomp14 ; where coperol=16285 and year>1997; var year matvest matunvest opts_vested_num_excl_curryear opts_vested_val_excl_curryear opts_unvested_num_excl_curryear opts_unvested_val_excl_curryear Xvest Xunvest Zvest Zunvest Svest Sunvest sumrealizable_value Rvest Runvest Vvest Vunvest; title "checking vested and unvested option values for ibm"; run; proc print data=excomp14 (obs=100); var coperol year delta ropt shrown; *where coperol=16285 and year>1997; where coperol=2611; run; proc sort data=excomp14; by year; proc means noprint data=excomp14; by year; var delta ropt ; output out=out4 mean = delta ropt ; proc print data=out4; title "checking means of calculated delta and vega by year"; run; data chk8; set excomp14; where (delta=. and ropt^=.); proc means; var shrown; title "appears that when ropt is not missing but delta is missing, it is because share ownership is missing"; title2 "2 obs only have shrownpc when delta is missing, ropt is not missing"; run; data chk9; set excomp14; where (delta^=. and ropt=.); title "appears that when delta is not missing, ropt is not missing"; run; **/ data excomp15; set excomp14; * dropping these variables because in the next step, we are merging back with full anncomp dataset that has these variables already; keep coperol year delta optiondelta sharedelta ropt Vportfolio_yearend exec_fullname; run; * now merge back the anncomp full dataset = recall that we have been working with a subset to make the program faster; proc sql; create table excomp16 as select * from excomp15 x left join anncomp y on x.coperol=y.coperol and x.year=y.year; /** *===============================; * READ DATA ON PLAN BASED AWARDS; *===============================; * data from 2006 onwards; * has no duplicates by coperol-year-grntnum; * WRDS says that earlier, grantnum referred to the grant number; * that is, it uniquely identified each option grant award; * now it identifies any award - stock, option,equity, and non-equity incentive plan awards; * WE DO NOT USE THIS DATA ANYWHERE IN THIS PROGRAM - SEE NOTES IN MAIN PAPER AS TO WHY; * WE HAVE A SEPARATE PROGRAM THAT VALUES THESE AWARDS (EIP_VALUE.SAS); data planbasedawards; set common.planbasedawards; coperol=co_per_rol; * round off the expric since this is sometimes rounded off in outstanding table but not in the planbased table, which creates problems with the merge; expric = round(expric,0.01); keep gvkey coperol year grntnum act_date eq_max eq_targ eq_thres expric fair_value grant_date mktpric non_eq_max non_eq_targ non_eq_thres opts_grt shares_grt; proc sort nodupkey; by coperol year grntnum; *proc contents; * title "information on dataset planbasedawards"; run; proc print data=planbasedawards; var coperol year grntnum mktpric expric grant_date opts_grt shares_grt; where coperol=16285; title "plan based awards table for ibm samuel palmisano"; run; * the planbased awards are options/stocks/other awards during the year; * if opts_grt>0, we take it as an straight option award; * if shares_grt>0, we take it as a straight share award; * we will use B-S value of option grants as on grant date for TDC1 adjustment * we do not need this for the delta and vega and wealth as these are determined as on fiscal year end and are based on options in the outstanding awards table; * similarly, for stock grants, we will use the grant date value for TDC1 adjustment; * For delta, vega and wealth, we use the shrown_excl_opts and price as of fiscal year end; * if opt_grts=0 and shares_grt=0 , then it is a long term performance award (equity or non-equity); * within long term performance award, it appears that some firms give options that are earned on meeting specific performance criteria, while most give shares; * if opts_grt=. and shares_grt=. and non-equity=. but there is an equity award and exercise price given, we assume that it is a performance award (EIP) with options; * we need to value these using B-S value of the options assuming target level payouts; * if opts_grt=. and shares_grt=. and non-equity=. and exercise price=. but there is an equity award we assume that it is a performance award (EIP) with shares; * we need to value these using grant date stock price and assuming target level payouts; * if opts_grt=. and shares_grt=. and equity=. but non-equity is not missing, then we assume that it is a non-equity performance award (NEIP); * we do not use the target/threshold etc. values of these as non-equity awards are given in ANNCOMP (NONEQ_INCENT) based on actual payouts; * to calculate B-S value as on grant date, we need exdate; * but this is given only in the outstanding awards table; * so we try and match the two tables to pull the exdate from the outstanding awards table; * I am going to create a subset of planbasedawards so that we don't match on missing values; * 190 executibve years appear to have expric missing; * cannot do anything about this anyway; * going to indicate whether there is only 1 award per record or multiple awards per record; * this will affect the fair value calculation - when there are multiple awards per record (see example elsewhere in program, the fair value represents the cumulative fair value of all the awards. So we are trying to see how big a problem this is; data planbasedawards1; attrib granttype length=$14.; set planbasedawards; if opts_grt=. and shares_grt=. and (eq_max=. and eq_targ=. and eq_thres=.) and (non_eq_max^=. or non_eq_targ^=. or non_eq_thres^=.) then granttype = "NEIP ONLY"; if opts_grt=. and shares_grt^=. and (eq_max=. and eq_targ=. and eq_thres=.) and (non_eq_max=. and non_eq_targ=. and non_eq_thres=.) then granttype = "PURE STOCK"; else if opts_grt^=. and shares_grt=. and (eq_max=. and eq_targ=. and eq_thres=.) and (non_eq_max=. and non_eq_targ=. and non_eq_thres=.) then granttype = "PURE OPTIONS"; else if opts_grt=. and shares_grt=. and (eq_max^=. or eq_targ^=. or eq_thres^=.) and expric^=. and (non_eq_max=. and non_eq_targ=. and non_eq_thres=.) then granttype = "EIP OPTIONS"; else if opts_grt=. and shares_grt=. and (eq_max^=. or eq_targ^=. or eq_thres^=.) and expric=. and (non_eq_max=. and non_eq_targ=. and non_eq_thres=.) then granttype = "EIP STOCK"; else if opts_grt=. and shares_grt=. and (eq_max=. and eq_targ=. and eq_thres=.) and (non_eq_max=. and non_eq_targ=. and non_eq_thres=.) then granttype = "ALL MISSING"; else granttype = "MULTIPLE"; run; proc print data=planbasedawards1; * var coperol year eq_max eq_targ eq_thres expric non_eq_max non_eq_targ non_eq_thres; * where opts_grt=. and shares_grt=. and (eq_max^=. or eq_targ^=. or eq_thres^=.) and expric^=.; where coperol=11664; * where coperol=16285; run; proc freq data=planbasedawards1; tables granttype; title "frequency of different types of plan based awards"; run; **/ *================================; * READ DATA ON OUTSTANDING AWARDS; *================================; * this dataset has individual executive awards outstanding at end of fiscal year; * should have no duplicates by coperol, award number (outawdnum),and year; * data only available after 2006; * appears to have a lot of the same information as in coperol; data outstandingawards; set execcomp.outstandingawards; coperol=co_per_rol; opts_vested_num = opts_unex_exer; opts_unvested_num = opts_unex_unexer; opts_unearned_num = opts_unex_unearn ; * round off the expric since this is sometimes rounded off in outstanding talbe but not in the planbased table, which creates problems with the merge; expric = round(expric,0.01); keep coperol year eip_shrs_unvest_num eip_shrs_unvest_val expric opts_vested_num opts_unvested_num opts_unearned_num outawdnum shrs_unvest_num shrs_unvest_val exdate; proc sort nodupkey; by coperol year outawdnum; /** proc contents data=common.outstandingawards; title "information on dataset outstandingawards"; run; **/ /** proc print data=outstandingawards; where coperol=16285; var year eip_shrs_unvest_num eip_shrs_unvest_val expric opts_vested_num opts_unvested_num opts_unearned_num outawdnum shrs_unvest_num shrs_unvest_val exdate; title "printing ibm post-2006 data on outstanding awards"; title2 "ibm changed reporting format in 2006 itself"; run; **/ proc means data=outstandingawards noprint n mean p1 p25 p50 p75 p99; var opts_unearned_num; where opts_unearned_num>0 and opts_unearned_num^=.; run; * we merge with outstanding awards to find the delta, vega, and value as of fiscal year end; data excomp13_subset; set excomp13; keep coperol year shrown bs_yield fyenddt sigmasq sigma prccf year oneyr twoyr threeyr fouryr fiveyr sixyr sevenyr eightyr nineyr tenyr; proc sort nodupkey; by coperol year; run; * now we need to compute the deltas of the portfolio as of year end; * outstandingawards does not have the data to compute risk-free rate etc, so we merge these in from excomp12_subset; proc sql; create table outstandingawards1 as select * from outstandingawards x left join excomp13_subset y on x.coperol=y.coperol and x.year=y.year; *proc contents data=outstandingawards1; *run; data outstandingawards1a; set outstandingawards1; *if we want a 70% haircut, we need to apply it here for post-2006; maturity_yearend = (exdate-fyenddt)/365; if round(maturity_yearend,1) =1 then rfyearend=oneyr; if round(maturity_yearend,1) =2 then rfyearend=twoyr; if round(maturity_yearend,1) =3 then rfyearend=threeyr; if round(maturity_yearend,1) =4 then rfyearend=fouryr; if round(maturity_yearend,1) =5 then rfyearend=fiveyr; if round(maturity_yearend,1) =6 then rfyearend=sixyr; if round(maturity_yearend,1) =7 then rfyearend=sevenyr; if round(maturity_yearend,1) =8 then rfyearend=eightyr; if round(maturity_yearend,1) =9 then rfyearend=nineyr; if round(maturity_yearend,1)=10 then rfyearend=tenyr; * sometimes maturity is greater than 10; if (maturity_yearend^=. and round(maturity_yearend,1)>10) then rfyearend=tenyr; Xc = expric; * calculate B-S values; Zc_yearend= (log(prccf/Xc)+ maturity_yearend*(rfyearend-bs_yield+sigmasq/2))/(sigma*sqrt( maturity_yearend)); if maturity_yearend=. then Zc_yearend=.; * computing deltas; Sunvest_yearend= exp(-bs_yield*maturity_yearend)*probnorm(Zc_yearend)*opts_unvested_num*prccf/100; if Zc_yearend=. then Sunvest_yearend=.; if opts_unvested_num=. or opts_unvested_num=0 then Sunvest_yearend=0; Svest_yearend= exp(-bs_yield*maturity_yearend)*probnorm(Zc_yearend)*opts_vested_num*prccf/100; if Zc_yearend=. then Svest_yearend=.; if opts_vested_num=. or opts_vested_num=0 then Svest_yearend=0; * Black Scholes value of options at yearend; Vunvest_yearend= opts_unvested_num*(prccf*exp(-bs_yield* maturity_yearend)*probnorm(Zc_yearend) -Xc*exp(-rfyearend*maturity_yearend)*probnorm(Zc_yearend-sigma*sqrt( maturity_yearend))); if Zc_yearend=. then Vunvest_yearend=.; if opts_unvested_num=. or opts_unvested_num=0 then Vunvest_yearend=0; Vvest_yearend= opts_vested_num*(prccf*exp(-bs_yield* maturity_yearend)*probnorm(Zc_yearend) -Xc*exp(-rfyearend*maturity_yearend)*probnorm(Zc_yearend-sigma*sqrt( maturity_yearend))); if Zc_yearend=. then Vvest_yearend=.; if opts_vested_num=. or opts_vested_num=0 then Vvest_yearend=0; * sensitivity with respect to a 0.01 change in stock return volatility; * see core and guay appendix A; if Zc_yearend^=. then do; Runvest_yearend = exp(-bs_yield* maturity_yearend)*PDF('normal',Zc_yearend,0,1)*prccf*sqrt( maturity_yearend)*0.01*opts_unvested_num; end; if Zc_yearend=. or maturity_yearend=. then Runvest_yearend=.; if opts_unvested_num=. or opts_unvested_num=0 then Runvest_yearend=0; if Zc_yearend^=. then do; Rvest_yearend = exp(-bs_yield* maturity_yearend)*PDF('normal',Zc_yearend,0,1)*prccf*sqrt( maturity_yearend)*0.01*opts_vested_num; end; if Zc_yearend=. or maturity_yearend=. then Rvest_yearend=.; if opts_vested_num=. or opts_vested_num=0 then Rvest_yearend=0; drop oneyr twoyr threeyr fouryr fiveyr sixyr sevenyr eightyr nineyr tenyr ; run; /** proc print data=outstandingawards1a; where coperol=16285 and year>1997; var year outawdnum Zc_yearend Sunvest_yearend Runvest_yearend Vunvest_yearend Svest_yearend Rvest_yearend Vvest_yearend maturity_yearend; title "calculating post-2006 numbers for ibm palmisano"; run; proc print data=outstandingawards1a; where coperol=668 and year=2009 and maturity_yearend^=.; var year sigma rfyearend bs_yield maturity_yearend; title "print of BS estimates for Molson Coors"; title2 "per proxy, company uses 28% volatility, 2.46% risk free rate, 2.29% dividend yield, and 5-7 year maturity"; run; **/ * now aggregate from tranche level to coperol-year level; proc sql; create table outstandingawards2 as select *, sum(opts_unvested_num) as sumunvested_num_opts, sum(opts_vested_num) as sumvested_num_opts, sum(Sunvest_yearend) as sumSunvest_yearend, sum(Svest_yearend) as sumSvest_yearend, sum(Runvest_yearend) as sumRunvest_yearend, sum(Rvest_yearend) as sumRvest_yearend, sum(Vunvest_yearend) as sumVunvest_yearend, sum(Vvest_yearend) as sumVvest_yearend from outstandingawards1a group by coperol,year; data outstandingawards3; set outstandingawards2; if sumunvested_num_opts=0 or sumunvested_num_opts=. then do; sumSunvest_yearend=0; sumRunvest_yearend=0; sumVunvest_yearend=0; end; if sumvested_num_opts=0 then do; sumSvest_yearend=0; sumRvest_yearend=0; sumVvest_yearend=0; end; * vega; newRopt = sum(sumRvest_yearend, sumRunvest_yearend); * sensitivity of shareholdings; Sshr= shrown*prccf/100; * value of shareholdings; newVshr= shrown*prccf; newVportfolio_yearend = sum(sumVvest_yearend, sumVunvest_yearend, newVshr); * delta; newdelta = sum(sumSvest_yearend, sumSunvest_yearend, Sshr); newoptiondelta = sum(sumSvest_yearend, sumSunvest_yearend); newsharedelta = Sshr; drop opts_vested_num opts_unvested_num shrown; * at this point, we are going to keep only 1 observation per executive year; * since we have already added up the values of current year's grants, that is all we need to keep; * the following sort elimates 290,425 obs; * these are executives reporting multiple stock grants outstanding at the end of each year; proc sort nodupkey ; by coperol year; run; /** proc print data=outstandingawards3; where coperol=16285; var year newdelta newRopt sumSunvest_yearend sumSvest_yearend sumRunvest_yearend sumRvest_yearend sumVunvest_yearend sumVvest_yearend newVshr newVportfolio_yearend; title "calculating post-2006 numbers for ibm palmisano"; run; **/ * MERGING THEM TOGETHER; proc sql; create table excomp17 as select * from excomp16 x left join outstandingawards3 y on x.coperol=y.coperol and x.year=y.year; data excomp18; set excomp17; if old_datafmt_flag=0 then do; delta=newdelta; optiondelta=newoptiondelta; sharedelta=newsharedelta; Ropt = newRopt; Vportfolio_yearend = newVportfolio_yearend; end; firm_related_wealth=Vportfolio_yearend; vega=ropt; keep coperol year delta vega firm_related_wealth old_datafmt_flag tdc1 gvkey exec_fullname sharedelta optiondelta; * keep gvkey coperol year delta vega firm_related_wealth ; proc sort nodupkey; by coperol year; /** proc means n mean median min max; title "Final dataset"; run; **/ /** proc print data=excomp18; var coperol year firm_related_wealth; where coperol=984; title "printing for GE -WELCH to check with Sundaram-Yermack table 1"; run; proc sort data=excomp18; by year; proc means noprint data=excomp18; by year; var delta vega firm_related_wealth ; output out=out1 mean=delta vega firm_related_wealth n=ndelta nvega nfirm_related_wealth ; run; proc print data=out1; title "Key variables"; run; **/ proc sort data=excomp18; by year coperol; run; data deltavega; set excomp18; keep gvkey coperol year delta vega firm_related_wealth sharedelta optiondelta; run; /* data deltavega; set deltavega; if optiondelta=0 then optiondelta=.; if delta=0 then delta=.; run; */ proc download data=deltavega out=local.deltavega; run; endrsubmit; signoff; |

Hey Kai,

First of all, thanks for sharing your SAS-script, it’s very useful!

As I aim to gather the Delta’s and Vega’s of the CEOs of companies in the S&P 1500 during 2007 – 2014, I hope you can help me solving a question that is related to the output of the script (I want to match the variables Delta and Vega to my dataset of which the GVKEY is the primary identifier);

– The output shows multiple rows of data per company per year. I assume this is because it also includes data on other executives, besides the CEO. I wish to collect only the data of CEOs. Do you think this is possible?

Best regards,

Thomas

You are right – the output includes other executives than CEO. You can modify the program to only include CEO. Another way (may be easier) is to SQL query if CO_PER_ROL is linked to a CEO flag in Execucomp.

By CEO flag, do you mean by ‘ceoann’ or ‘pceo’? I know this is so basic, but I never used this data set before and the variable explanation of these two provided by WRDS is not very clear to me, it would be so great if you can help to clarify this.

I think annCEO would do the trick

Hi, Kai,

In the following part:

”

if fyr=6 then assumed_grantyear=year-1;

else assumed_grantyear=year;

assumed_grantdate=mdy(7,1,assumed_grantyear);

”

should it be “fyr<=6" instead of "fyr=6", since options granted in the first 5 months should also be assigned the previous year as the grant year?

Bo has informed me that Naveen’s program is correct—Naveen uses the code because of the way Compustat defines data year. Naveen shows the example in details in the program. Thank you Bo for letting me know this.

Hi Kai,

Excellent code. If you can I think it might be valuable to add in ‘moneyness’ calculations too:

Campbell, T.C., Gallmeyer, M., Johnson, S.A., Rutherford, J., Stanley, B.W., 2011. CEO

optimism and forced turnover. J. Financ. Econ. 101, 695-712.

Hi Kai,

Thank you very much for this excellent code.

I am not able to find these two variables: fybegdt fyenddt. I am using Stata so I need to download them.

Are they supposed to be available on CRSP/Compustat Merged Database – Security Monthly?

Many thanks for your help.

The first several lines in the program generates fybegdt and fyenddt. They are not something that can be directly downloaded from WRDS. You need SAS to run the program. Stata will not do the work.

Hi Kai:

Thank you very much for the codes. However, the coperol from the download data, I assume, is modified and ranked by the order of firms in the sample, instead of the true co_per_rol? How could I solve it? I checked the codes but couldn’t figure out where the definition of co_per_rol has been changed.

I don’t quite understand your question. If I remember correctly, co_per_rol is the unique id of an executive, no matter which company he/she works for. The code won’t change co_per_rol throughout.

Nevermind. I figured it out. It’s due to the sorting in the SAS data.

Hi Kai,

Thanks for sharing the codes! I’m a bit confused about the following codes at the end of the program.

data deltavega;

set deltavega;

if optiondelta=0 then optiondelta=.;

if delta=0 then delta=.;

run;

You corrected Naveen’s codes by using sum function. Sum function itself returns missing value if all variables summed up are missing. If we set zero optiondelta and delta to missing, aren’t we dropping some observations with legitimate value zero?

Hi Emily, thanks for letting me know. I think you’re correct. I cannot recall exactly why I added these codes. Probably because I misunderstood the sum function. I commented them out. Glad I had that disclaimer from day 1 🙂

Hi Chen:

I am wondering if you could upload the codes for total annual compensation (Execucomp variable TDC1)adjustment, as suggested by Coles, Daniel, and Naveen (2014, RFS). They adjust the total compensation for the changes in reporting following FAS 123R and new SECdisclosure requirements. I think Naveen already mentioned something about the adjustment in her codes on calculating delta and vega, but the information is very limited.

Hi Kai,

Thanks for your excellent code! I am a bit confused by the following parts:

A. line 1030

data excomp13;

set excomp12;

if sumnumsecur=0 or sumnumsecur=. then do;

Sopts_grants_yearend=0;

Vopts_grantdate=0;

Ropts_grants_yearend=0;

sumrealizable_value=0;

Vopts_yearend=0;

FVopts_yearend=0;

end;

where does this FVopts_yearend come from?

B. line 1289

if opts_unvested_val_excl_curryear=. or opts_unvested_val_excl_curryear<0

or opts_unvested_num_excl_curryear1) then rfunvest=tenyr;

It should be if (matunvest^=. and round(matunvest,1)>10) then rfunvest=tenyr; right?

D. line 1330

if (matvest^=. and round(matvest,10)>10) then rfvest=tenyr;

It should be if (matvest^=. and round(matvest,1)>10) then rfvest=tenyr;.

Thanks for your reply!

Hi Kai,

Hi Kai,

Thanks for your excellent code!

Sorry for the typos in the previous comment, so I re-send it again.

I am a bit confused by the following parts:

A. line 1030

data excomp13;

set excomp12;

if sumnumsecur=0 or sumnumsecur=. then do;

Sopts_grants_yearend=0;

Vopts_grantdate=0;

Ropts_grants_yearend=0;

sumrealizable_value=0;

Vopts_yearend=0;

FVopts_yearend=0;

end;

where does this FVopts_yearend come from?

B. line 1288

if opts_unvested_val_excl_curryear=. or opts_unvested_val_excl_curryear<0

or opts_unvested_num_excl_curryear1) then rfunvest=tenyr;

It should be if (matunvest^=. and round(matunvest,1)>10) then rfunvest=tenyr; right?

D. line 1330

if (matvest^=. and round(matvest,10)>10) then rfvest=tenyr;

It should be if (matvest^=. and round(matvest,1)>10) then rfvest=tenyr;.

Thanks for your reply!

Hi Emma:

Do you mean line 1274

if (matunvest^=. and round(matunvest,1)>1) then rfunvest=tenyr;

should be

if (matunvest^=. and round(matunvest,1)>10) then rfunvest=tenyr;

I think Naveen mess up with the round function, all of these two round functions should be round(matvest,1)>10.

FVopts_yearend=0, I have no idea about this because this variable only shows up one time.

Hi Kai,

Thank you very much for this great code 🙂

I am wondering if I can get delta for the current year’s option grants after 2006. In your code, delta for new options is “Sopts_grants_yearend”, however, I cannot find the similar variable for “newdelta” part (after 2006).

Thank you!

Could you run his code? I couldn’t find the risk free rates dataset in wrds. The library name frb.rates_daily.

I have problem with the library name. Libraries like crsp.ccmxpf_lnkhist and crsp.msf turn out to be error in my sas program. When I try to assign library name, I found that crsp in crsp.ccmxpf_lnkhist is /wrds/crsp/sasdata/a_ccm. But crsp in crsp.msf is /wrds/crsp/sasdata/a_stock. Hence, do I need to rename every library manually. However, it seems that your script could be run directly. So how should i do it?

Dear Kai! Thank you very much for such a helpful website! The code is great! I was trying to use the data of vega/delta to replicate simultaneous equations with R&D and CAPEX in Coles, J., Daniel, N., Naveen, L., 2006, but I get different results. Any chance that you tried to do it as well based on your data? Kind regards, Anastasia

No. I didn’t try to replicate the results of this paper. Generally, it is extremely hard to replicate the results of other research. It sounds not right but it is truth. Different results can occur at many data processing steps, e.g., how and when the researcher winsorizes variables? how and when the researcher drops certain observations. This is a black box. That’s why I think our community should increase the transparency of coding.

yes, I absolutely agree with you! Thank you again, Kai! and I’ll be always happy to see any new posts from you.

Dear Kai, thank you for posting the code! I wanted to kindly ask you for a clarification. Vvest – stands for value of vested options, while “Vunvest” – as a value for unvested options. Can you, please explain the meaning of “Vopts_yearend” (that is the year-coperol sum of “Vc_yearend”)

+ +

This is actually less of a question for Kai than it is for Lalitha Naveen and her coauthors, they note that this is a variable that they create for their Pay-for-luck paper (see line 910), which is forthcoming in RFS (Symmetry in Pay for Luck).

However, the answer is reasonably clear from the code (see lines 882-913):

– ‘Vc_yearend’ is the year-end-Black-Scholes value for the securities in each option grant.

– ‘Vopts_yearend’ is the year-end-Black-Scholes value for all the options granted in the year. Line 952 simply sums the year-end BS values across all option grants in the year. This is an input to the BS value of the executive’s firm related equity portfolio on line 1395.

I know it’s kind of stupid to ask ,

but do you need a remote user name and remote password to access at the very beginning.

I am a PHD student I can only access to Compustat via the link our school provide so which means I can not use the way you provide right?

If i could able to download the data from Compustat, could you please tell me where should I start??

Many many thanks.

Lucas,

There are three ways to run this code (the links provide more information):

(1) From a computer with SAS installed

http://wrds-www.wharton.upenn.edu/pages/support/programming-wrds/programming-sas/sas-from-your-computer/

(2) On the WRDS Cloud

http://wrds-www.wharton.upenn.edu/pages/support/programming-wrds/programming-sas/sas-wrds-cloud/

(3) On the web via the WRDS website

http://wrds-www.wharton.upenn.edu/pages/support/programming-wrds/programming-sas/sas-web-sasstudio/

Kai’s excellent code is written to run using method (1) with minor modifications (i.e. line 8 should be changed to match a location on your machine). When you run the code it will prompt you for the credentials you use to login to the WRDS website.

Method (2) also requires the credentials you use to log on to the WRDS site, but with several additional modifications. Follow the link above if you’re interested.

Method (3) does not require you to have SAS on your local computer.

If you have access to these datasets through your university, then you should be able to execute Kai’s code using one of these methods. I’d start with method (3) if it’s your first time as it has the fewest requirements.

Best,

Arthur

Dear Arthur

May I ask for some help? I am using the method 3 here.

I basically copied and pasted the entire code into the web browser, I deleted anything above

rsubmit;

I added

%include ‘/wrds/lib/utility/wrdslib.sas’ ;

options sasautos=(‘/wrds/wrdsmacros/’, SASAUTOS) MAUTOSOURCE;

But the code will not run. I do not know what is wrong.

Hey Kai,

thanks for this really great code, that made work much easier!! However, when I want to continue working with data on option, and I even want to compare options before (option_awards_blk_value) and after 2006….which variable to use for the post period (option_awards_fv option_or awards?) or do I even need further calculations?

Many thanks again.

Best, Uli

Hi Kai,

thanks for your great code, that really made work easier! However, when I want to continue to work with option data and need to compare data from 2003 to 2008, which variables do I have to use? For the pre period “option_awards_blk_value” but for the post period (one of these “option_awards” or “fv option_awards”) or do I even need further calculations?

Many thanks again.

Best,

Uli

Uli,

The answer to your question has two parts.

First, Lalitha Naveen’s original code (and Kai’s excellent update) both address the changes in the structure of option data on Execucomp around 2006 to come as close as is probably practical to a consistent option granting/holdings/delta/vega time series. For more details on exactly how Kai and Lalitha address these changes it’s worth reading the annotations in the code in detail, as well as the working paper that Jeff Coles, Naveen Daniel, and Lalitha Naveen have on SSRN: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2296381.

Second, comparing option data from 2003 to option data from 2008 should be done with great care, even if your data is perfect. Remember that in the pre-2006 (pre-SFAS 123R) options were expensed at their intrinsic value (most often zero) rather than their fair value (greater than zero). Now options are expensed at their fair values, this accounting change seems to have had a huge effect on the use of option grants in compensation.

Some papers that offer a good discussion of these changes are:

Mary Ellen Carter, Luann J. Lynch, and I˙rem Tuna (2007) The Role of Accounting in the Design of CEO Equity Compensation. The Accounting Review: March 2007, Vol. 82, No. 2, pp. 327-357.

Bettis, J.C., Bizjak, J., Coles, J.L. and Kalpathy, S., 2018. Performance-vesting provisions in executive compensation. Journal of Accounting and Economics, 66(1), pp.194-221.

Core, John E. and Packard, Heidi, Non-price and Price Performance Vesting Provisions and Executive Incentives (July 28, 2017). Available at SSRN: https://ssrn.com/abstract=2547590 or http://dx.doi.org/10.2139/ssrn.2547590

Best,

Arthur

Many thanks. Arthur!!

Hi Kai and everyone,

Did you manage to calculate the Delta and Vega for the years 2015-2016-2017?

Naveen and Coles do not report them on their website, therefore we cannot make sure that they are correct without calculating the previous years.

Thank you!

Jules

Jules,

Kai’s excellent code does this easily.

You need to change line 67 from:

%let end_year=2014;

to:

%let end_year=2018;

or whatever the relevant year is.

I believe Kai set line 67 to 2014 for comparability to Lalitha’s code.

Best,

Arthur

Hello Kai, Arther, and Everyone,

Thank you so much for this great website. Could you kindly answer the following three questions:

(Q.1) Alike total delta incentives and total vega incentives, is it possible to get annual delta (i.e., delta grant) and annual vega (i.e., vega grant) incentives?

I am sure you are familiar with this, but I would like to get these variables that appear in Core and Guay 1999.

(Q.2) In addition to the interest rates that appear in the following link,

https://wrds-www.wharton.upenn.edu/pages/support/manuals-and-overviews/compustat/execucomp/modified-black-scholes-option-valuation-methodology/

where can I find the rates for post 2003 (to 2012)?

(Q.3) To compute PPS of each executive following Edmans, Gabaix, and Landier (2009), the formula seems to be: PPS = [# of shares + # of options * option delta) * (price / 100) / annual compensation

Thanks to you codes, I believe I got the ‘option delta’. Would it be possible to get the option vega?

Also, I wish I knew the variable name for “price”, but perhaps this is not something that should be asked here.

Hi Kai,

Thank you for sharing your code!

I do have a question though: by comparing your delta with Naveen’s version, I find the difference becomes significantly larger after 2007. This makes me think whether whatever causes the increasing difference would induce more issues when I extend the sample period to recent years.

Could you help me out on this? Thank you!

Best regards,

Jason

Hi Kai,

Thank you for the great code!!

Just to be sure I understand: the units on delta, vega, and firm-related wealth are thousands of dollars in contemporaneous prices, correct?

Thanks again!

Dear Kai,

I’m using your wonderful code, and would like to appropriately cite you.

You write that we should cite Naveen’s work when using your code (and her website makes clear which papers she thinks should be cited), but only write “I would be appreciated if you are generous enough to acknowledge my work.”

Is there a paper of yours that would be most appropriate for me to cite in reference to this code? At the moment, I have a “thank you” to you, but please let me know if you’d like a different citation.

Thanks again for doing this,

Dave

Hi Kai,

Thank you for sharing your code!

I appreciate your work but i need to up date my data till 2019. I will be gratuful if you can send the new data upadated because i don’t have now access to Execucomp and CRSP.

Many thanks

Kaou

Dear Dr Chen

May I ask

For line 417

I received the following warning

WARNING: Variable GVKEY already exists on file WORK.EXCOMP1

Is this normal?

For line 1628-1631 I received the following:

ERROR: Neither the PRINT option nor a valid output statement has been given.

What is wrong?

Hi Joey,

I might be helpful.

(1) Line 417. This is a typical warning when you use proc SQL to join two datasets when each of them has the same variable, say GVKEY here. Here, SAS reads in the first GVKEY from the first dataset and then the second GVKEY in the second dataset. Since the resulting dataset cannot have two identical variables so the SAS does not keep the second GVKEY and gives a warning. You will see “already exists” in multiple similar cases in this code elsewhere. No worry.

(2) Line 1628-1631. This is just an error recognized by the program. But actually, this is not an error in the calculation since line 1628-1631 is designed to do a basic checking based on quantiles. Prof. Kai Chen inserts “noprint” to Prof. Naveen’s code to prevent printing results in SAS since printing results takes time. This waste of time is not necessary for executing the right calculation. No worry.

Thanks Li, I am not familiar with SAS, so your help is extremely helpful

Dear Dr. Chun,

Thank you so much for your your code to calculate delta and vega

ExecComp and Compustat fundamental data both have the same libname (comp) and this makes problem, which stops running the SAS program. Do you know how to distinguish two?

libname comp /*remote*/ ‘/wrds/comp/sasdata/d_na’ /*server=wrds*/;

libname comp /*remote*/ ‘/wrds/comp/sasdata/execcomp’ /*server=wrds*/;

Also I see execcomp.person, execcomp.ex_black in the program, but I do not see them in ExecComp database. Could you please explain this?

Thank you so much for your help

Dear Prof. Chen,

I spot a tiny typo in line 1274. The 1 should be 10 since the condition is for var matunvest to be bigger than 10.

Old and typoed code:

if (matunvest^=. and round(matunvest,1)>1) then rfunvest=tenyr;

New and correct code:

if (matunvest^=. and round(matunvest,1)>10) then rfunvest=tenyr;

Hi Prof. Chen

I just see a code error in line 1274, where the code shall assign frunvest=tenyr when the matunvest is greater than 10.

The old and errored code is:

if (matunvest^=. and round(matunvest,1)>1) then rfunvest=tenyr;

The new and corrected code shall be:

if (matunvest^=. and round(matunvest,1)>10 then rfunvest=tenyr;

I also go to Prof. Naveen’s website to check. Her code is correct in the corresponding line (line 1135) at https://sites.temple.edu/lnaveen/files/2020/11/deltavega_2013.txt.

Best Regards,

Bo Li

Hi Prof. Chen

I just see a code error in line 1339, where the code shall assign frunvest=tenyr when the matunvest is greater than 10. I also check with Prof. Naveen. She said this error is corrected in her updated code, though her online version does not update this change when I communicated with her.

in Line 1339, the old and errored code is:

if (matvest^=. and round(matvest,10)>10) then rfvest=tenyr;

The new and corrected code shall be:

if (matvest^=. and round(matvest,1)>10) then rfvest=tenyr;

The reason is actually quite simple. The old code can leave rfvest unassigned (missing value assigned by SAS) for observations that have matvest between 10.50 and 14.99. Such observations will then have missing value for Zvest, Svest, Rvest, and Vvest.

I further write a short piece of code and find out that there are 172 observations in excomp14 that have rfvest as missing value due to above reason. After above correction, out of these 172 observations, 152 observations have non-missing value for all four variables Zvest, Svest, Rvest, and Vvest. This is not a big issue at all given the huge number of total observations. But it is good to correct it.

Thank you again for making this code available. I definitely learn a lot from such code.

Best Regards,

Bo Li