Comparison of all 90 numbered tables in Empirical Research in Accounting: Tools and Methods (R edition). Page numbers refer to printed folios.
45 tables have numerical changes; 38 retain the same displayed numbers; 7 have no numerical entries.
The comparison ignores digit grouping, signed zero and changes in row order when the corresponding numbers agree. Significance annotations were also checked for tables with unchanged numbers. Each changed table below includes a cause assessment and a check of its interpretation. Controlled checks hold current inputs fixed; unresolved differences are identified explicitly. See the audit explanation.
Cause: Code correction; isolated. On identical saved simulation draws, restoring sd(x)/sqrt(n-1) reproduces all three printed rates (0.0508, 0.0967, 0.1161); sd(x)/sqrt(n) gives the online rates (0.0511, 0.0971, 0.1166). No external data are involved. Source correction.
Effect on interpretation: Unchanged: the direct test is close to 5%; the alternative and choice rules over-reject.
Example: Rejection rate: prop_sig_diff: 0.0508 in the hard-copy edition; 0.0511 online.
Null hypothesis rejection rates for different approaches
prop_sig_diff prop_sig_diff_alt prop_sig_diff_choice
0.0508 0.0967 0.1161Table 5.1: Null hypothesis rejection rates for different approaches
| prop_sig_diff | prop_sig_diff_alt | prop_sig_diff_choice |
|---|---|---|
| 0.0511 | 0.0971 | 0.1166 |
Cause: Updated data. The printed and current count queries use the same grouping variables and no date restriction. The underlying Compustat counts have changed.
Effect on interpretation: Unchanged: INDL/C/D/STD remains the dominant combination.
Example: INDL/C/D/STD observations: 573,430 in the hard-copy edition; 599,379 online.
Observations by (indfmt, consol, popsrc, datafmt) values
indfmt consol popsrc datafmt n
INDL C D STD 573,430
INDL C D SUMM_STD 279,283
FS C D STD 46,814
INDL P D STD 1,171
INDL R D STD 480
INDL P D SUMM_STD 300
FS R D STD 51
INDL R D SUMM_STD 28
INDL D D SUMM_STD 25
INDL D D STD 25
INDL C D PRE_AMENDS 19
INDL C D PRE_AMENDSS 10Table 6.1: Observations by (indfmt, consol, popsrc, datafmt) values
| indfmt | consol | popsrc | datafmt | n |
|---|---|---|---|---|
| INDL | C | D | STD | 599,379 |
| INDL | C | D | SUMM_STD | 290,465 |
| FS | C | D | STD | 48,716 |
| INDL | P | D | STD | 1,179 |
| INDL | R | D | STD | 480 |
| INDL | P | D | SUMM_STD | 306 |
| FS | R | D | STD | 51 |
| INDL | R | D | SUMM_STD | 28 |
| INDL | D | D | SUMM_STD | 25 |
| INDL | D | D | STD | 25 |
| INDL | C | D | PRE_AMENDS | 19 |
| INDL | C | D | PRE_AMENDSS | 10 |
Cause: Updated data. The relevant linking and counting rules agree with the printed code. CRSP/Compustat inputs have changed; formatting changes and an explicit by= argument do not change these rules.
Effect on interpretation: Unchanged: most links agree, but neither link dates nor CUSIP links are interchangeable in every case. Table 7.9 now includes two cases with the same PERMNO and different PERMCOs; the prose does not assert that this is impossible.
Example: NR links: 43,255 in the hard-copy edition; 46,853 online.
Distribution of linktype in crsp.ccmxpf_lnkhist
has_permno linktype n
FALSE NR 43,255
FALSE NU 34,125
FALSE NP 4
TRUE LC 17,013
TRUE LU 15,993
TRUE LS 5,158Table 7.3: Distribution of linktype in crsp.ccmxpf_lnkhist
| has_permno | linktype | n |
|---|---|---|
| FALSE | NR | 46853 |
| FALSE | NU | 34014 |
| FALSE | NP | 4 |
| TRUE | LC | 17932 |
| TRUE | LU | 15945 |
| TRUE | LS | 7159 |
Cause: Updated data. The relevant linking and counting rules agree with the printed code. CRSP/Compustat inputs have changed; formatting changes and an explicit by= argument do not change these rules.
Effect on interpretation: Unchanged: most links agree, but neither link dates nor CUSIP links are interchangeable in every case. Table 7.9 now includes two cases with the same PERMNO and different PERMCOs; the prose does not assert that this is impossible.
Example: P links: 29,689 in the hard-copy edition; 32,292 online.
Distribution of linkprim
for valid links
linkprim n
P 29,689
C 7,961
J 420
N 94Table 7.6: Distribution of linkprim for valid links
| linkprim | n |
|---|---|
| P | 32292 |
| C | 8190 |
| J | 452 |
| N | 102 |
Cause: Updated data. The relevant linking and counting rules agree with the printed code. CRSP/Compustat inputs have changed; formatting changes and an explicit by= argument do not change these rules.
Effect on interpretation: Unchanged: most links agree, but neither link dates nor CUSIP links are interchangeable in every case. Table 7.9 now includes two cases with the same PERMNO and different PERMCOs; the prose does not assert that this is impossible.
Example: GVKEYs with one PERMNO: 32,198 in the hard-copy edition; 34,628 online.
Distribution of number of PERMNOs
matching each GVKEY
num_permnos n
1 32,198
2 1,056
3 76
4 3
5 1Table 7.7: Distribution of number of PERMNOs matching each GVKEY
| num_permnos | n |
|---|---|
| 1 | 34628 |
| 2 | 1214 |
| 3 | 76 |
| 4 | 5 |
| 5 | 1 |
Cause: Updated data. The relevant linking and counting rules agree with the printed code. CRSP/Compustat inputs have changed; formatting changes and an explicit by= argument do not change these rules.
Effect on interpretation: Unchanged: most links agree, but neither link dates nor CUSIP links are interchangeable in every case. Table 7.9 now includes two cases with the same PERMNO and different PERMCOs; the prose does not assert that this is impossible.
Example: Same PERMNO and PERMCO at both dates: 124,673 in the hard-copy edition; 135,207 online.
Distribution of CRSP-Compustat links at datadate versus rdq same_permno same_permco has_permno_rdq has_permno_ddate n TRUE TRUE TRUE TRUE 124,673 FALSE FALSE FALSE TRUE 4,577 FALSE FALSE TRUE FALSE 3,059 FALSE FALSE TRUE TRUE 24 FALSE TRUE TRUE TRUE 22
Table 7.8: Distribution of CRSP-Compustat links at datadate versus rdq
| same_permno | same_permco | has_permno_rdq | has_permno_ddate | n |
|---|---|---|---|---|
| TRUE | TRUE | TRUE | TRUE | 135207 |
| FALSE | FALSE | FALSE | TRUE | 4300 |
| FALSE | FALSE | TRUE | FALSE | 3570 |
| FALSE | FALSE | TRUE | TRUE | 31 |
| FALSE | TRUE | TRUE | TRUE | 27 |
Cause: Updated data. The relevant linking and counting rules agree with the printed code. CRSP/Compustat inputs have changed; formatting changes and an explicit by= argument do not change these rules.
Effect on interpretation: Unchanged: most links agree, but neither link dates nor CUSIP links are interchangeable in every case. Table 7.9 now includes two cases with the same PERMNO and different PERMCOs; the prose does not assert that this is impossible.
Example: Same PERMNO and PERMCO under both links: 242,840 in the hard-copy edition; 255,521 online.
Comparison of CCM- and CUSIP-based links
same_permno same_permco n
FALSE FALSE 131
FALSE TRUE 160
TRUE TRUE 242,840
NA NA 118,522Table 7.9: Comparison of CCM- and CUSIP-based links
| same_permno | same_permco | n |
|---|---|---|
| FALSE | FALSE | 100 |
| FALSE | TRUE | 169 |
| TRUE | FALSE | 2 |
| TRUE | TRUE | 255521 |
| NA | NA | 125518 |
Cause: Updated data. The relevant linking and counting rules agree with the printed code. CRSP/Compustat inputs have changed; formatting changes and an explicit by= argument do not change these rules.
Effect on interpretation: Unchanged: most links agree, but neither link dates nor CUSIP links are interchangeable in every case. Table 7.9 now includes two cases with the same PERMNO and different PERMCOs; the prose does not assert that this is impossible.
Example: Covered by both link methods: 243,131 in the hard-copy edition; 255,792 online.
Open current online Table 7.10
Differences in coverage of CCM- and CUSIP-based link tables
has_permno_ccm has_permno_cusip n
FALSE TRUE 3,233
TRUE FALSE 115,289
TRUE TRUE 243,131Table 7.10: Differences in coverage of CCM- and CUSIP-based link tables
| has_permno_ccm | has_permno_cusip | n |
|---|---|---|
| FALSE | TRUE | 3270 |
| TRUE | FALSE | 122248 |
| TRUE | TRUE | 255792 |
Cause: Updated data. The Apple regressions retain the same specifications and an open-ended CRSP sample; observations rise from 4,781 to 5,033.
Effect on interpretation: The positive event-volume association is smaller (0.060 to 0.036), but remains significant at 1%. The signs and significance of the other displayed slopes are unchanged.
Example: Observations in each regression: 4,781 in the hard-copy edition; 5,033 online.
Open current online Table 13.1
Regression analysis of Apple media events
ret_mkt abs(ret) Volume
(Intercept) 0.001*** 0.014*** 0.997***
is_eventTRUE −0.002** 0.001 0.060
Num.Obs. 4,781 4,781 4,781
R2 0.001 0.000 0.000Table 13.1: Regression analysis of Apple media events
| ret_mkt | abs(ret) | Volume | |
|---|---|---|---|
| (Intercept) | 0.001*** | 0.014*** | 0.999*** |
| is_eventTRUE | -0.002** | 0.001 | 0.036 |
| Num.Obs. | 5033 | 5033 | 5033 |
| R2 | 0.001 | 0.000 | 0.000 |
Cause: Updated data. The fiscal-year/date query and both regression formulas are unchanged; the Compustat input has been refreshed.
Effect on interpretation: Unchanged: fiscal year is closely related to calendar year and fiscal-end month. This is an illustrative date convention, not a causal estimate.
Example: Intercept, column (2): −1.124 in the hard-copy edition; −1.059 online.
Open current online Table 14.2
Regression of fyear on month indicators
(1) (2)
factor(month)1 −1.000
factor(month)2 −1.000
factor(month)3 −1.000
factor(month)4 −1.000
factor(month)5 −1.000
factor(month)6 0.000
factor(month)7 0.000
factor(month)8 0.000
factor(month)9 0.000
factor(month)10 0.000
factor(month)11 0.000
factor(month)12 0.000
year 1.000 1.000
(Intercept) −1.124
month 0.122
R2 1.000 1.000Table 14.2: Regression of fyear on month indicators
| (1) | (2) | |
|---|---|---|
| factor(month)1 | -1.000 | |
| factor(month)2 | -1.000 | |
| factor(month)3 | -1.000 | |
| factor(month)4 | -1.000 | |
| factor(month)5 | -1.000 | |
| factor(month)6 | -0.000 | |
| factor(month)7 | -0.000 | |
| factor(month)8 | -0.000 | |
| factor(month)9 | -0.000 | |
| factor(month)10 | -0.000 | |
| factor(month)11 | -0.000 | |
| factor(month)12 | -0.000 | |
| year | 1.000 | 1.000 |
| (Intercept) | -1.059 | |
| month | 0.122 | |
| R2 | 1.000 | 1.000 |
Cause: Code correction demonstrated; residual input effects unresolved. Running the printed return-window code and corrected code on the same current inputs changes the persistence-regression N from 12,831 to 41,643. The printed table has 12,877. Corrected calendar arithmetic restores month-end observations, and the benchmark window is twelve months. Ordering before ntile() also makes ties reproducible. The residual print-versus-old-code difference cannot be assigned solely to data without the original input snapshot. Source correction.
Effect on interpretation: The accrual-persistence and negative accrual-return conclusions remain. Some magnitudes change substantially. In Table 15.10 the cash-flow coefficient gains only 10% significance; the text makes no contrary claim. The accrual hedge return remains positive and significant, with a smaller magnitude.
Example: Mean accruals, decile 1: −0.1751 in the hard-copy edition; −0.1688 online.
Open current online Table 15.2
Means by accruals decile
acc_decile acc earn cfo
1 −0.1751 0.0355 0.2106
2 −0.0871 0.0946 0.1817
3 −0.0614 0.1035 0.1649
4 −0.0442 0.1109 0.1551
5 −0.0302 0.1178 0.1481
6 −0.0168 0.1182 0.1351
7 −0.0006 0.1269 0.1275
8 0.0226 0.1353 0.1127
9 0.0599 0.1422 0.0823
10 0.1724 0.1539 −0.0185Table 15.2: Means by accruals decile
| acc_decile | acc | earn | cfo |
|---|---|---|---|
| 1 | -0.1688 | 0.0516 | 0.2204 |
| 2 | -0.0869 | 0.0968 | 0.1838 |
| 3 | -0.0614 | 0.1077 | 0.1691 |
| 4 | -0.0443 | 0.1111 | 0.1554 |
| 5 | -0.0303 | 0.1133 | 0.1437 |
| 6 | -0.0171 | 0.1158 | 0.1329 |
| 7 | -0.0007 | 0.1273 | 0.1280 |
| 8 | 0.0224 | 0.1404 | 0.1179 |
| 9 | 0.0593 | 0.1492 | 0.0899 |
| 10 | 0.1718 | 0.1582 | -0.0136 |
Cause: Code correction demonstrated; residual input effects unresolved. Running the printed return-window code and corrected code on the same current inputs changes the persistence-regression N from 12,831 to 41,643. The printed table has 12,877. Corrected calendar arithmetic restores month-end observations, and the benchmark window is twelve months. Ordering before ntile() also makes ties reproducible. The residual print-versus-old-code difference cannot be assigned solely to data without the original input snapshot. Source correction.
Effect on interpretation: The accrual-persistence and negative accrual-return conclusions remain. Some magnitudes change substantially. In Table 15.10 the cash-flow coefficient gains only 10% significance; the text makes no contrary claim. The accrual hedge return remains positive and significant, with a smaller magnitude.
Example: Observations in each regression: 12,877 in the hard-copy edition; 41,643 online.
Open current online Table 15.3
Replication of Table 2 of Sloan (1996)
(1) (2)
(Intercept) 0.030*** 1.396***
(0.001) (0.039)
earn 0.704***
(0.005)
earn_decile 0.749***
(0.006)
Num.Obs. 12,877 12,877
R2 0.607 0.553Table 15.3: Replication of Table 2 of Sloan ( 1996 )
| (1) | (2) | |
|---|---|---|
| (Intercept) | 0.026*** | 1.197*** |
| (0.000) | (0.020) | |
| earn | 0.746*** | |
| (0.003) | ||
| earn_decile | 0.782*** | |
| (0.003) | ||
| Num.Obs. | 41643 | 41643 |
| R2 | 0.610 | 0.601 |
Cause: Code correction demonstrated; residual input effects unresolved. Running the printed return-window code and corrected code on the same current inputs changes the persistence-regression N from 12,831 to 41,643. The printed table has 12,877. Corrected calendar arithmetic restores month-end observations, and the benchmark window is twelve months. Ordering before ntile() also makes ties reproducible. The residual print-versus-old-code difference cannot be assigned solely to data without the original input snapshot. Source correction.
Effect on interpretation: The accrual-persistence and negative accrual-return conclusions remain. Some magnitudes change substantially. In Table 15.10 the cash-flow coefficient gains only 10% significance; the text makes no contrary claim. The accrual hedge return remains positive and significant, with a smaller magnitude.
Example: Mean industry-level earn coefficient: 0.664 in the hard-copy edition; 0.749 online.
Open current online Table 15.4
Industry-level persistence with actual earnings
term mean q1 median q3
(Intercept) 0.026 0.012 0.026 0.036
earn 0.664 0.611 0.697 0.815Table 15.4: Industry-level persistence with actual earnings
| term | mean | q1 | median | q3 |
|---|---|---|---|---|
| (Intercept) | 0.023 | 0.015 | 0.022 | 0.031 |
| earn | 0.749 | 0.685 | 0.742 | 0.823 |
Cause: Code correction demonstrated; residual input effects unresolved. Running the printed return-window code and corrected code on the same current inputs changes the persistence-regression N from 12,831 to 41,643. The printed table has 12,877. Corrected calendar arithmetic restores month-end observations, and the benchmark window is twelve months. Ordering before ntile() also makes ties reproducible. The residual print-versus-old-code difference cannot be assigned solely to data without the original input snapshot. Source correction.
Effect on interpretation: The accrual-persistence and negative accrual-return conclusions remain. Some magnitudes change substantially. In Table 15.10 the cash-flow coefficient gains only 10% significance; the text makes no contrary claim. The accrual hedge return remains positive and significant, with a smaller magnitude.
Example: Mean industry-level earn_decile coefficient: 0.670 in the hard-copy edition; 0.755 online.
Open current online Table 15.5
Industry-level persistence with earnings deciles
term mean q1 median q3
(Intercept) 1.612 0.937 1.257 1.745
earn_decile 0.670 0.649 0.739 0.819
Table 15.5: Industry-level persistence with earnings deciles
| term | mean | q1 | median | q3 |
|---|---|---|---|---|
| (Intercept) | 1.306 | 0.879 | 1.253 | 1.528 |
| earn_decile | 0.755 | 0.710 | 0.759 | 0.827 |
Cause: Code correction demonstrated; residual input effects unresolved. Running the printed return-window code and corrected code on the same current inputs changes the persistence-regression N from 12,831 to 41,643. The printed table has 12,877. Corrected calendar arithmetic restores month-end observations, and the benchmark window is twelve months. Ordering before ntile() also makes ties reproducible. The residual print-versus-old-code difference cannot be assigned solely to data without the original input snapshot. Source correction.
Effect on interpretation: The accrual-persistence and negative accrual-return conclusions remain. Some magnitudes change substantially. In Table 15.10 the cash-flow coefficient gains only 10% significance; the text makes no contrary claim. The accrual hedge return remains positive and significant, with a smaller magnitude.
Example: Observations in each regression: 12,877 in the hard-copy edition; 41,643 online.
Open current online Table 15.6
Replication of Table 3 of Sloan (1996)
(1) (2)
(Intercept) 0.028*** −1.312***
(0.001) (0.073)
acc 0.640***
(0.007)
cfo 0.723***
(0.005)
acc_decile 0.534***
(0.007)
cfo_decile 0.760***
(0.007)
Num.Obs. 12,877 12,877
R2 0.613 0.470Table 15.6: Replication of Table 3 of Sloan ( 1996 )
| (1) | (2) | |
|---|---|---|
| (Intercept) | 0.024*** | -1.624*** |
| (0.000) | (0.038) | |
| acc | 0.706*** | |
| (0.004) | ||
| cfo | 0.757*** | |
| (0.003) | ||
| acc_decile | 0.550*** | |
| (0.004) | ||
| cfo_decile | 0.792*** | |
| (0.004) | ||
| Num.Obs. | 41643 | 41643 |
| R2 | 0.612 | 0.520 |
Cause: Code correction demonstrated; residual input effects unresolved. Running the printed return-window code and corrected code on the same current inputs changes the persistence-regression N from 12,831 to 41,643. The printed table has 12,877. Corrected calendar arithmetic restores month-end observations, and the benchmark window is twelve months. Ordering before ntile() also makes ties reproducible. The residual print-versus-old-code difference cannot be assigned solely to data without the original input snapshot. Source correction.
Effect on interpretation: The accrual-persistence and negative accrual-return conclusions remain. Some magnitudes change substantially. In Table 15.10 the cash-flow coefficient gains only 10% significance; the text makes no contrary claim. The accrual hedge return remains positive and significant, with a smaller magnitude.
Example: Mean industry-level acc coefficient: 0.6708 in the hard-copy edition; 0.7133 online.
Open current online Table 15.7
Industry-level persistence with earnings components
term mean q1 median q3
(Intercept) 0.0247 0.0098 0.0239 0.0315
acc 0.6708 0.5895 0.6669 0.7659
cfo 0.7135 0.6514 0.7228 0.8276
Table 15.7: Industry-level persistence with earnings components
| term | mean | q1 | median | q3 |
|---|---|---|---|---|
| (Intercept) | 0.0210 | 0.0129 | 0.0214 | 0.0276 |
| acc | 0.7133 | 0.6319 | 0.7032 | 0.7795 |
| cfo | 0.7691 | 0.7091 | 0.7585 | 0.8467 |
Cause: Code correction demonstrated; residual input effects unresolved. Running the printed return-window code and corrected code on the same current inputs changes the persistence-regression N from 12,831 to 41,643. The printed table has 12,877. Corrected calendar arithmetic restores month-end observations, and the benchmark window is twelve months. Ordering before ntile() also makes ties reproducible. The residual print-versus-old-code difference cannot be assigned solely to data without the original input snapshot. Source correction.
Effect on interpretation: The accrual-persistence and negative accrual-return conclusions remain. Some magnitudes change substantially. In Table 15.10 the cash-flow coefficient gains only 10% significance; the text makes no contrary claim. The accrual hedge return remains positive and significant, with a smaller magnitude.
Example: Mean industry-level acc_decile coefficient: 0.4854 in the hard-copy edition; 0.5300 online.
Open current online Table 15.8
Industry-level persistence with earnings component deciles
term mean q1 median q3
(Intercept) −0.8800 −2.0286 −1.2842 −0.6100
acc_decile 0.4854 0.4376 0.5201 0.5982
cfo_decile 0.6966 0.6724 0.7353 0.8455Table 15.8: Industry-level persistence with earnings component deciles
| term | mean | q1 | median | q3 |
|---|---|---|---|---|
| (Intercept) | -1.6624 | -2.3245 | -1.5626 | -1.1088 |
| acc_decile | 0.5300 | 0.4705 | 0.5195 | 0.5835 |
| cfo_decile | 0.7989 | 0.7385 | 0.7968 | 0.8602 |
Cause: Code correction demonstrated; residual input effects unresolved. Running the printed return-window code and corrected code on the same current inputs changes the persistence-regression N from 12,831 to 41,643. The printed table has 12,877. Corrected calendar arithmetic restores month-end observations, and the benchmark window is twelve months. Ordering before ntile() also makes ties reproducible. The residual print-versus-old-code difference cannot be assigned solely to data without the original input snapshot. Source correction.
Effect on interpretation: The accrual-persistence and negative accrual-return conclusions remain. Some magnitudes change substantially. In Table 15.10 the cash-flow coefficient gains only 10% significance; the text makes no contrary claim. The accrual hedge return remains positive and significant, with a smaller magnitude.
Example: Observations in each regression: 12,867 in the hard-copy edition; 41,623 online.
Open current online Table 15.9
Abnormal returns, earnings, and lagged components
(1) (2)
(Intercept) −0.027*** 0.186***
(0.006) (0.018)
lead_earn 2.538***
(0.061)
acc −1.961***
(0.062)
cfo −1.826***
(0.057)
lead_earn_decile 0.082***
(0.002)
acc_decile −0.055***
(0.002)
cfo_decile −0.056***
(0.002)
Num.Obs. 12,867 12,867
R2 0.124 0.109
Table 15.9: Abnormal returns, earnings, and lagged components
| (1) | (2) | |
|---|---|---|
| (Intercept) | -0.038*** | 0.160*** |
| (0.003) | (0.009) | |
| lead_earn | 2.443*** | |
| (0.033) | ||
| acc | -1.939*** | |
| (0.035) | ||
| cfo | -1.823*** | |
| (0.032) | ||
| lead_earn_decile | 0.073*** | |
| (0.001) | ||
| acc_decile | -0.047*** | |
| (0.001) | ||
| cfo_decile | -0.054*** | |
| (0.001) | ||
| Num.Obs. | 41623 | 41623 |
| R2 | 0.118 | 0.097 |
Cause: Code correction demonstrated; residual input effects unresolved. Running the printed return-window code and corrected code on the same current inputs changes the persistence-regression N from 12,831 to 41,643. The printed table has 12,877. Corrected calendar arithmetic restores month-end observations, and the benchmark window is twelve months. Ordering before ntile() also makes ties reproducible. The residual print-versus-old-code difference cannot be assigned solely to data without the original input snapshot. Source correction.
Effect on interpretation: The accrual-persistence and negative accrual-return conclusions remain. Some magnitudes change substantially. In Table 15.10 the cash-flow coefficient gains only 10% significance; the text makes no contrary claim. The accrual hedge return remains positive and significant, with a smaller magnitude.
Example: Observations in each regression: 13,798 in the hard-copy edition; 44,488 online.
Open current online Table 15.10
Regressions for Abel and Mishkin (1983) test
(1) (2)
(Intercept) 0.041*** 0.078***
(0.006) (0.018)
acc −0.327***
(0.049)
cfo 0.038
(0.037)
acc_decile −0.011***
(0.002)
cfo_decile 0.006***
(0.002)
Num.Obs. 13,798 13,798
R2 0.005 0.007Table 15.10: Regressions for Abel and Mishkin ( 1983 ) test
| (1) | (2) | |
|---|---|---|
| (Intercept) | 0.023*** | 0.046*** |
| (0.003) | (0.009) | |
| acc | -0.217*** | |
| (0.027) | ||
| cfo | 0.038* | |
| (0.021) | ||
| acc_decile | -0.007*** | |
| (0.001) | ||
| cfo_decile | 0.004*** | |
| (0.001) | ||
| Num.Obs. | 44488 | 44488 |
| R2 | 0.003 | 0.004 |
Cause: Code correction demonstrated; residual input effects unresolved. Running the printed return-window code and corrected code on the same current inputs changes the persistence-regression N from 12,831 to 41,643. The printed table has 12,877. Corrected calendar arithmetic restores month-end observations, and the benchmark window is twelve months. Ordering before ntile() also makes ties reproducible. The residual print-versus-old-code difference cannot be assigned solely to data without the original input snapshot. Source correction.
Effect on interpretation: The accrual-persistence and negative accrual-return conclusions remain. Some magnitudes change substantially. In Table 15.10 the cash-flow coefficient gains only 10% significance; the text makes no contrary claim. The accrual hedge return remains positive and significant, with a smaller magnitude.
Example: acc_decile1 coefficient: 0.120*** in the hard-copy edition; 0.077*** online.
Open current online Table 15.11
Partial replication of Table 6 of Sloan
(1996)
(1)
acc_decile1 0.120***
(0.020)
acc_decile2 0.109***
(0.020)
acc_decile3 0.042**
(0.020)
acc_decile4 0.058***
(0.020)
acc_decile5 0.067***
(0.020)
acc_decile6 0.042**
(0.020)
acc_decile7 0.025
(0.020)
acc_decile8 0.034*
(0.020)
acc_decile9 −0.003
(0.020)
acc_decile10 −0.012
(0.020)
Num.Obs. 300
R2 0.255
Table 15.11: Partial replication of Table 6 of Sloan ( 1996 )
| (1) | |
|---|---|
| acc_decile1 | 0.077*** |
| (0.011) | |
| acc_decile2 | 0.077*** |
| (0.011) | |
| acc_decile3 | 0.046*** |
| (0.011) | |
| acc_decile4 | 0.033*** |
| (0.011) | |
| acc_decile5 | 0.024** |
| (0.011) | |
| acc_decile6 | 0.019* |
| (0.011) | |
| acc_decile7 | 0.026** |
| (0.011) | |
| acc_decile8 | 0.019* |
| (0.011) | |
| acc_decile9 | -0.009 |
| (0.011) | |
| acc_decile10 | -0.007 |
| (0.011) | |
| Num.Obs. | 300 |
| R2 | 0.320 |
Cause: Code corrections plus input/sample changes. Eligibility and event-year selection were corrected to retain enough estimation observations, and the Industry Model now uses industry-year rather than pooled industry medians. The selected firms/years consequently change. A full numerical split from database revisions is unavailable. Source correction.
Effect on interpretation: Some test-size inferences change: for Healy neg_p05 the binomial p-value is 0.0817 rather than 0.0071; Healy pos_p05 is 0.9422 rather than 0.0088. Conversely, DeAngelo neg_p01 and pos_p01 now reject correct size at 5% (p-values 0.0020 and 0.0165, both 0.1860 in print). Jones and Modified Jones still over-reject. The chapter now makes clear that missizing differs by model and tail.
Example: Mean DeAngelo estimate: 0.0124 in the hard-copy edition; 0.0061 online.
Open current online Table 16.1
Results of tests of earning management: Sample 1
measure stat mean sd q1 median q3
deangelo estimate 0.0124 0.3083 −0.0778 −0.0036 0.0676
deangelo statistic −0.0071 1.1151 −0.6277 −0.0297 0.6169
deangelo std.error 0.1841 0.3048 0.0781 0.1298 0.2030
healy estimate 0.0104 0.3820 −0.0604 −0.0062 0.0477
healy statistic 0.0589 1.3072 −0.6509 −0.0869 0.6303
healy std.error 0.1287 0.2330 0.0556 0.0909 0.1413
industry estimate 0.0099 0.3780 −0.0601 −0.0061 0.0474
industry statistic 0.0577 1.3189 −0.6520 −0.0758 0.6252
industry std.error 0.1275 0.2308 0.0553 0.0903 0.1402
jones estimate 0.0142 0.3664 −0.0498 −0.0014 0.0550
jones statistic 0.0421 1.9776 −0.7897 −0.0256 0.7928
jones std.error 0.0890 0.0942 0.0444 0.0699 0.1039
mod_jones estimate 0.0108 0.3755 −0.0531 −0.0020 0.0535
mod_jones statistic 0.0163 1.8795 −0.8040 −0.0307 0.7861
mod_jones std.error 0.0918 0.0968 0.0456 0.0715 0.1089Table 16.1: Results of tests of earnings management: Sample 1
| measure | stat | mean | sd | q1 | median | q3 |
|---|---|---|---|---|---|---|
| deangelo | estimate | 0.0061 | 0.1725 | -0.0410 | 0.0050 | 0.0565 |
| deangelo | statistic | 0.0699 | 1.1236 | -0.4534 | 0.0589 | 0.6145 |
| deangelo | std.error | 0.1241 | 0.0977 | 0.0605 | 0.0998 | 0.1541 |
| healy | estimate | 0.0028 | 0.1026 | -0.0379 | 0.0010 | 0.0376 |
| healy | statistic | 0.0656 | 1.0640 | -0.5234 | 0.0230 | 0.6780 |
| healy | std.error | 0.0900 | 0.0700 | 0.0428 | 0.0716 | 0.1137 |
| industry | estimate | 0.0016 | 0.1125 | -0.0393 | 0.0024 | 0.0401 |
| industry | statistic | 0.0488 | 1.7719 | -0.5853 | 0.0459 | 0.7618 |
| industry | std.error | 0.0808 | 0.0650 | 0.0373 | 0.0627 | 0.1028 |
| jones | estimate | 0.0028 | 0.1015 | -0.0354 | 0.0023 | 0.0363 |
| jones | statistic | 0.0534 | 1.3144 | -0.6261 | 0.0506 | 0.7521 |
| jones | std.error | 0.0707 | 0.0524 | 0.0368 | 0.0581 | 0.0891 |
| mod_jones | estimate | 0.0043 | 0.1038 | -0.0355 | 0.0025 | 0.0403 |
| mod_jones | statistic | 0.0672 | 1.3253 | -0.6060 | 0.0591 | 0.7776 |
| mod_jones | std.error | 0.0732 | 0.0544 | 0.0367 | 0.0602 | 0.0921 |
Cause: Code corrections plus input/sample changes. Eligibility and event-year selection were corrected to retain enough estimation observations, and the Industry Model now uses industry-year rather than pooled industry medians. The selected firms/years consequently change. A full numerical split from database revisions is unavailable. Source correction.
Effect on interpretation: Some test-size inferences change: for Healy neg_p05 the binomial p-value is 0.0817 rather than 0.0071; Healy pos_p05 is 0.9422 rather than 0.0088. Conversely, DeAngelo neg_p01 and pos_p01 now reject correct size at 5% (p-values 0.0020 and 0.0165, both 0.1860 in print). Jones and Modified Jones still over-reject. The chapter now makes clear that missizing differs by model and tail.
Example: Jones neg_p05 rejection rate: 0.0860 in the hard-copy edition; 0.0670 online.
Open current online Table 16.2
Type I error rates
measure neg_p01 neg_p05 pos_p01 pos_p05
deangelo 0.0141 0.0489 0.0141 0.0522
healy 0.0100 0.0320 0.0290 0.0690
industry 0.0100 0.0320 0.0300 0.0670
jones 0.0320 0.0860 0.0500 0.1000
mod_jones 0.0310 0.0870 0.0490 0.1010Table 16.2: Type I error rates
| measure | tests | neg_p01 | neg_p05 | pos_p01 | pos_p05 |
|---|---|---|---|---|---|
| deangelo | 1000.0000 | 0.0210 | 0.0550 | 0.0180 | 0.0580 |
| healy | 1000.0000 | 0.0120 | 0.0380 | 0.0180 | 0.0490 |
| industry | 989.0000 | 0.0222 | 0.0556 | 0.0364 | 0.0758 |
| jones | 1000.0000 | 0.0220 | 0.0670 | 0.0270 | 0.0770 |
| mod_jones | 1000.0000 | 0.0250 | 0.0670 | 0.0270 | 0.0790 |
Cause: Code corrections plus input/sample changes. Eligibility and event-year selection were corrected to retain enough estimation observations, and the Industry Model now uses industry-year rather than pooled industry medians. The selected firms/years consequently change. A full numerical split from database revisions is unavailable. Source correction.
Effect on interpretation: Some test-size inferences change: for Healy neg_p05 the binomial p-value is 0.0817 rather than 0.0071; Healy pos_p05 is 0.9422 rather than 0.0088. Conversely, DeAngelo neg_p01 and pos_p01 now reject correct size at 5% (p-values 0.0020 and 0.0165, both 0.1860 in print). Jones and Modified Jones still over-reject. The chapter now makes clear that missizing differs by model and tail.
Example: Healy neg_p05 p-value: 0.0071 in the hard-copy edition; 0.0817 online.
Open current online Table 16.3
p-values for null that Type I error rates equal size of tests
measure neg_p01 neg_p05 pos_p01 pos_p05
deangelo 0.1860 0.9398 0.1860 0.7619
healy 1.0000 0.0071 0.0000 0.0088
industry 1.0000 0.0071 0.0000 0.0165
jones 0.0000 0.0000 0.0000 0.0000
mod_jones 0.0000 0.0000 0.0000 0.0000Table 16.3: p-values for null that Type I error rates equal size of tests
| measure | neg_p01 | neg_p05 | pos_p01 | pos_p05 |
|---|---|---|---|---|
| deangelo | 0.0020 | 0.4676 | 0.0165 | 0.2452 |
| healy | 0.5215 | 0.0817 | 0.0165 | 0.9422 |
| industry | 0.0006 | 0.4216 | 0.0000 | 0.0004 |
| jones | 0.0007 | 0.0165 | 0.0000 | 0.0003 |
| mod_jones | 0.0000 | 0.0165 | 0.0000 | 0.0001 |
Cause: Code correction; sample window. The placebo illustration now starts in Q1 2001, matching the stated period, instead of Q1 2000. This is a deliberate sample restriction, not an extra year of refreshed data.
Effect on interpretation: Unchanged: the placebo estimate remains negative and significant; the effect is -0.317 instead of -0.343 and the t-statistic is -4.62 instead of -4.96. The numerical commentary already agrees with the corrected table.
Example: Treatment × post coefficient: −0.343*** in the hard-copy edition; −0.317*** online.
Open current online Table 19.2
Test of parallel trends in bid-ask spreads
(1)
(Intercept) 2.830***
(0.049)
postTRUE −2.130***
(0.044)
treatTRUE 0.489***
(0.077)
postTRUE × treatTRUE −0.343***
(0.069)
Num.Obs. 220,822Table 19.2: Test of parallel trends in bid-ask spreads
| (1) | |
|---|---|
| (Intercept) | 2.457*** |
| (0.052) | |
| postTRUE | -1.757*** |
| (0.046) | |
| treatTRUE | 0.465*** |
| (0.077) | |
| postTRUE × treatTRUE | -0.317*** |
| (0.069) | |
| Num.Obs. | 186696 |
Cause: Code/specification corrections plus input changes. The accrual construction and FHK replication specifications were corrected, including firm and calendar-year fixed effects. Table 19.4 now displays the two treatment interactions and uses the printed t-based RI procedure. These changes cannot be reduced to a data refresh.
Effect on interpretation: Inference is sensitive to specification and method: during-treatment p-values in Table 19.4 are 0.0449 (firm clustering), 0.0398 (two-way clustering), and 0.0507 (RI). Post-treatment p-values exceed 0.28. A blanket claim of significance across all three methods would be incorrect; the text makes no such claim.
Example: Observations, column (4): 19,616 in the hard-copy edition; 21,804 online.
Open current online Table 19.3
Baseline replication of FHK
(1) (2) (3) (4)
pilotTRUE 0.000 0.000
(0.004) (0.004)
pilotTRUE × duringTRUE −0.010** −0.010** −0.014** −0.014**
(0.004) (0.004) (0.005) (0.005)
pilotTRUE × postTRUE 0.009** 0.008** 0.005 0.005
(0.004) (0.004) (0.004) (0.004)
Num.Obs. 19,616 19,616 19,616 19,616
Firm FEs No No Yes Yes
Controls No Yes No YesTable 19.3: Baseline replication of FHK
| (1) | (2) | (3) | (4) | |
|---|---|---|---|---|
| pilotTRUE | -0.002 | -0.002 | ||
| (0.004) | (0.004) | |||
| pilotTRUE × duringTRUE | -0.008* | -0.008* | -0.011** | -0.010** |
| (0.004) | (0.004) | (0.005) | (0.005) | |
| pilotTRUE × postTRUE | 0.010** | 0.009** | 0.006 | 0.006 |
| (0.004) | (0.004) | (0.005) | (0.006) | |
| Num.Obs. | 21816 | 21816 | 21804 | 21804 |
| Firm FEs | No | No | Yes | Yes |
| Controls | No | Yes | No | Yes |
Cause: Code/specification corrections plus input changes. The accrual construction and FHK replication specifications were corrected, including firm and calendar-year fixed effects. Table 19.4 now displays the two treatment interactions and uses the printed t-based RI procedure. These changes cannot be reduced to a data refresh.
Effect on interpretation: Inference is sensitive to specification and method: during-treatment p-values in Table 19.4 are 0.0449 (firm clustering), 0.0398 (two-way clustering), and 0.0507 (RI). Post-treatment p-values exceed 0.28. A blanket claim of significance across all three methods would be incorrect; the text makes no such claim.
Example: Post-treatment coefficient, RI: 0.008 in the hard-copy edition; 0.0059 online.
Open current online Table 19.4
Results with randomization inference
(1) (2) (3)
log(at) 0.001 0.001 0.001
(0.001) (0.001) (0.001)
mtob −0.001** −0.001 −0.001*
(0.000) (0.000) (0.000)
roa −0.009 −0.009 −0.009
(0.012) (0.016) (0.012)
leverage −0.016*** −0.016** −0.016***
(0.004) (0.006) (0.004)
pilotTRUE × duringTRUE −0.010** −0.010** −0.010*
(0.005) (0.004) (0.005)
pilotTRUE × postTRUE 0.008 0.008** 0.008
(0.005) (0.004) (0.006)
Num.Obs. 19,616 19,616 19,616
SEs CL-i CL-2 RITable 19.4: Results with randomization inference
| term | method | estimate | std_error | p_value |
|---|---|---|---|---|
| pilotTRUE:duringTRUE | CL-i | -0.0103 | 0.0051 | 0.0449 |
| pilotTRUE:duringTRUE | CL-2 | -0.0103 | 0.0050 | 0.0398 |
| pilotTRUE:duringTRUE | RI | -0.0103 | 0.0053 | 0.0507 |
| pilotTRUE:postTRUE | CL-i | 0.0059 | 0.0057 | 0.2996 |
| pilotTRUE:postTRUE | CL-2 | 0.0059 | 0.0055 | 0.2830 |
| pilotTRUE:postTRUE | RI | 0.0059 | 0.0058 | 0.3024 |
Cause: Simulation/reporting difference; precise cause unresolved. Only the displayed IV R-squared changes (0.035 to 0.034); all printed coefficients, standard errors and significance annotations agree. This is simulated data, so an external database refresh cannot explain it. The original fitted object and software environment have not been reconstructed.
Effect on interpretation: Unchanged: the coefficient comparison illustrating OLS and IV is identical. No claim in the text rests on the last displayed R-squared digit.
Example: IV R-squared: 0.035 in the hard-copy edition; 0.034 online.
Open current online Table 20.1
Regression results
OLS IV
(Intercept) −0.054* −0.054*
(0.031) (0.031)
X 0.184*** 0.190***
(0.031) (0.032)
Num.Obs. 1,000 1,000
R2 0.035 0.035Table 20.1: Regression results
| OLS | IV | |
|---|---|---|
| (Intercept) | -0.054* | -0.054* |
| (0.031) | (0.031) | |
| X | 0.184*** | 0.190*** |
| (0.031) | (0.032) | |
| Num.Obs. | 1000 | 1000 |
| R2 | 0.035 | 0.034 |
Cause: Software effect isolated; residual input difference unresolved. On the current input, disabling fixed-effect singleton removal gives 28,707 observations, while the current default gives 28,230 with the same post coefficient. The printed N is 28,745. Singleton removal therefore accounts for 477 observations in this controlled comparison; the remaining 38-observation difference is unresolved.
Effect on interpretation: Unchanged at displayed precision: coefficients, standard errors and stars agree with print. N and R-squared differ; the reported association and its limitations are unchanged.
Example: Observations, column (1): 28,745 in the hard-copy edition; 28,230 online.
Open current online Table 21.4
Effect of IDD adoption
(1) (2) (3) (4)
postTRUE 0.016** 0.024*** 0.014 0.019**
(0.008) (0.008) (0.009) (0.010)
Num.Obs. 28,745 24,569 3,617 3,164
R2 0.672 0.711 0.624 0.661Table 21.4: Effect of IDD adoption
| (1) | (2) | (3) | (4) | |
|---|---|---|---|---|
| postTRUE | 0.016** | 0.024*** | 0.014 | 0.019** |
| (0.008) | (0.008) | (0.009) | (0.010) | |
| Num.Obs. | 28230 | 23923 | 3611 | 3154 |
| R2 | 0.665 | 0.703 | 0.623 | 0.660 |
Cause: Input/software effects; exact split unresolved. The displayed event-time coefficients change slightly, while the regression specifications and reference period are retained. Input data, package data and estimation software cannot be separated using the preserved print evidence.
Effect on interpretation: The displayed significance annotations are unchanged. The discussion and plot description retain the same qualitative pattern; small coefficient updates do not overturn it.
Example: t − 5 coefficient, column (1): −0.005 in the hard-copy edition; −0.004 online.
Open current online Table 21.5
Effect of IDD adoption by year
(1) (2) (3) (4)
t−5 −0.005 0.001 −0.017 −0.009
(0.019) (0.021) (0.027) (0.029)
t−4 −0.017 −0.030 −0.026 −0.038
(0.022) (0.023) (0.025) (0.026)
t−3 −0.030 −0.029 −0.031 −0.031
(0.020) (0.020) (0.021) (0.022)
t−2 −0.008 −0.003 −0.014 −0.010
(0.018) (0.018) (0.019) (0.019)
t+0 −0.001 0.016 0.007 0.020
(0.015) (0.015) (0.016) (0.017)
t+1 0.009 0.025 0.021 0.031
(0.016) (0.017) (0.018) (0.020)
t+2 −0.007 0.010 0.009 0.017
(0.017) (0.018) (0.021) (0.023)
t+3 0.023 0.021 0.038 0.026
(0.019) (0.020) (0.025) (0.027)
t+4 0.028 0.039* 0.041 0.042
(0.019) (0.021) (0.027) (0.030)
t+5 0.044*** 0.048*** 0.046 0.039
(0.015) (0.016) (0.032) (0.036)Table 21.5: Effect of IDD adoption by year
| (1) | (2) | (3) | (4) | |
|---|---|---|---|---|
| t − 5 | -0.004 | 0.001 | -0.017 | -0.009 |
| (0.019) | (0.021) | (0.027) | (0.029) | |
| t − 4 | -0.018 | -0.031 | -0.026 | -0.038 |
| (0.022) | (0.023) | (0.025) | (0.026) | |
| t − 3 | -0.030 | -0.029 | -0.031 | -0.031 |
| (0.020) | (0.020) | (0.021) | (0.022) | |
| t − 2 | -0.009 | -0.003 | -0.014 | -0.010 |
| (0.018) | (0.018) | (0.019) | (0.019) | |
| t + 0 | -0.001 | 0.016 | 0.007 | 0.020 |
| (0.015) | (0.015) | (0.016) | (0.017) | |
| t + 1 | 0.009 | 0.025 | 0.021 | 0.031 |
| (0.016) | (0.017) | (0.018) | (0.020) | |
| t + 2 | -0.008 | 0.010 | 0.009 | 0.017 |
| (0.017) | (0.018) | (0.021) | (0.023) | |
| t + 3 | 0.023 | 0.021 | 0.038 | 0.026 |
| (0.019) | (0.020) | (0.025) | (0.027) | |
| t + 4 | 0.028 | 0.039* | 0.041 | 0.042 |
| (0.019) | (0.021) | (0.027) | (0.030) | |
| t + 5 | 0.043*** | 0.048*** | 0.046 | 0.039 |
| (0.015) | (0.016) | (0.032) | (0.036) |
Cause: Input/software effects; exact split unresolved. The risk-asymmetry regressions retain their formulas. Their inputs include refreshed CRSP/Compustat and corrected package data; fixed-effect sample handling may also differ. The chapter-level Iliev data correction alone does not establish the cause of every changed cell.
Effect on interpretation: Unchanged: the firm-effects coefficient remains negative, significant at 1%, and close to Bloomfield's -0.302. The separate local RDD discussion still finds no significant effect; Table 22.2 is not that local RDD.
Example: Treatment × post coefficient, column (2): −0.331*** in the hard-copy edition; −0.328*** online.
Open current online Table 22.2
Effect of SOX 404 on risk asymmetry
(1) (2)
(Intercept) 0.267***
(0.023)
I(treat * post) −0.181*** −0.331***
(0.066) (0.101)
Num.Obs. 1,849 1,849
R2 0.004 0.194
Column (2) includes firm and year fixed effectsTable 22.2: Effect of SOX 404 on risk asymmetry
| (1) | (2) | |
|---|---|---|
| Column (2) includes firm and year fixed effects | ||
| (Intercept) | 0.267*** | |
| (0.023) | ||
| I(treat * post) | -0.179*** | -0.328*** |
| (0.066) | (0.100) | |
| Num.Obs. | 1849 | 1848 |
| R2 | 0.004 | 0.191 |
Cause: Updated data. The filing inventory runs through the available SEC quarters rather than a fixed ending quarter. Additional filings change the counts, form ranking and percentages.
Effect on interpretation: Unchanged: form-type counts are descriptive. The 8-K percentage changes slightly and no inference in the text relies on its exact printed value.
Example: Form 4 filings: 8,962,439 in the hard-copy edition; 9,688,528 online.
Open current online Table 23.4
Most common form types on SEC EDGAR
form_type n
4 8,962,439
8-K 1,956,527
SC 13G/A 854,515
3 807,001
424B2 757,350
10-Q 683,426
6-K 523,774
497 511,773
SC 13G 450,673
D 410,707Table 23.4: Most common form types on SEC EDGAR
| form_type | n |
|---|---|
| 4 | 9,688,528 |
| 8-K | 2,091,740 |
| 424B2 | 1,112,353 |
| 3 | 880,094 |
| SC 13G/A | 867,478 |
| 10-Q | 718,221 |
| 6-K | 580,132 |
| 497 | 541,052 |
| D | 487,537 |
| SC 13G | 456,345 |
Cause: Updated data. The filing inventory runs through the available SEC quarters rather than a fixed ending quarter. Additional filings change the counts, form ranking and percentages.
Effect on interpretation: Unchanged: form-type counts are descriptive. The 8-K percentage changes slightly and no inference in the text relies on its exact printed value.
Example: Percentage of 8-K variants classified as 8-K: 95.852 in the hard-copy edition; 95.975 online.
Open current online Table 23.5
Number of 8-K filings by variant
form_type n percent
8-K 1,956,527 95.852
8-K/A 83,473 4.089
8-K12G3 637 0.031
8-K12B 335 0.016
8-K12G3/A 115 0.006
8-K15D5 77 0.004
8-K12B/A 37 0.002
8-K15D5/A 2 0.000Table 23.5: Number of 8-K filings by variant
| form_type | n | percent |
|---|---|---|
| 8-K | 2,091,740 | 95.975 |
| 8-K/A | 86,455 | 3.967 |
| 8-K12G3 | 653 | 0.030 |
| 8-K12B | 385 | 0.018 |
| 8-K12G3/A | 116 | 0.005 |
| 8-K15D5 | 79 | 0.004 |
| 8-K12B/A | 42 | 0.002 |
| 8-K15D5/A | 4 | 0.000 |
Cause: Code corrections plus updated data. A left join retains 10-Ks with no matching 8-K, assigns zero counts, and selects one CIK match deterministically. The SEC/financial-statement inputs have also changed. No original input snapshot is available for a full split. Source correction.
Effect on interpretation: Unchanged: filing complexity is positively associated with subsequent 8-K counts in OLS, firm-effects and Poisson specifications, with 1% significance. Magnitudes and sample sizes change.
Example: ln_grossfilesize coefficient, OLS: 1.786*** in the hard-copy edition; 1.968*** online.
Open current online Table 23.6
Financial statement complexity and voluntary disclosure
OLS Firm FE Pois
(Intercept) −19.613*** −0.826***
(−107.209) (−42.136)
ln_grossfilesize 1.786*** 1.396*** 0.186***
(135.856) (52.776) (132.757)
mkt_cap 0.000 0.000 0.000***
(−0.518) (0.051) (−3.731)
size 0.526*** 0.415*** 0.053***
(51.698) (12.339) (46.069)
roa −0.003 −0.027*** 0.000
(−0.852) (−2.579) (−0.722)
mtb 0.000* 0.000 0.000
(−1.826) (−1.503) (−1.517)
special_items 0.004 0.023** 0.000
(0.973) (2.566) (0.934)
Num.Obs. 106,988 106,988 106,988
*p <0.1, **p <0.05, ***p <0.01Table 23.6: Financial statement complexity and voluntary disclosure
| OLS | Firm FE | Pois | |
|---|---|---|---|
| * p < 0.1, ** p < 0.05, *** p < 0.01 | |||
| (Intercept) | -22.501*** | -1.355*** | |
| (-131.420) | (-68.322) | ||
| ln_grossfilesize | 1.968*** | 1.557*** | 0.217*** |
| (158.136) | (111.593) | (153.452) | |
| mkt_cap | -0.000 | -0.000 | -0.000*** |
| (-0.734) | (-0.168) | (-4.516) | |
| size | 0.522*** | 0.459*** | 0.055*** |
| (53.239) | (22.136) | (47.032) | |
| roa | -0.004 | -0.026** | -0.000 |
| (-1.123) | (-2.369) | (-0.907) | |
| mtb | -0.000* | -0.000 | -0.000 |
| (-1.737) | (-0.667) | (-1.415) | |
| special_items | 0.005 | 0.022** | 0.001 |
| (1.234) | (2.345) | (1.096) | |
| Num.Obs. | 111155 | 108850 | 111155 |
Cause: Code/sample corrections plus input and numerical effects. Lags are calculated before exclusions, consecutive years and known auditors are required, and the robust fit has an explicit seed. Input revisions and robust-estimator software may also contribute. Source correction.
Effect on interpretation: The Big N coefficient remains negative and significant at 1% in the robust specification (-0.008 to -0.010). The lesson about sensitivity to treatment of extreme values remains.
Example: big_nTRUE coefficient, robust regression: −0.008*** in the hard-copy edition; −0.010*** online.
Open current online Table 24.1
Results from various approaches to extreme values
Wins, w/ROA Raw Wins Trunc Cook’s D Robust
(Intercept) −0.031*** 0.262 −0.085*** −0.119*** −0.107*** −0.023***
(−8.934) (1.204) (−7.352) (−14.799) (−10.300) (−16.929)
big_nTRUE 0.000 −0.073 0.006 0.009** 0.020*** −0.008***
(0.264) (−1.078) (1.316) (2.614) (3.930) (−12.677)
roa 0.815***
(63.785)
cfo −0.786*** 0.799* 0.257*** 0.124*** 0.194*** −0.386***
(−37.425) (1.762) (8.932) (5.573) (5.237) (−144.226)
size 0.003*** −0.066* 0.004** 0.007*** 0.001 0.003***
(5.116) (−1.862) (2.626) (5.680) (0.685) (13.730)
lev −0.035** −0.072 −0.111*** −0.031*** −0.074*** −0.019***
(−2.641) (−0.344) (−5.089) (−4.478) (−4.161) (−22.438)
mtb −0.001*** −0.003 −0.007*** −0.006*** −0.001*** 0.000***
(−4.393) (−0.938) (−8.695) (−11.031) (−3.855) (−4.471)
Num.Obs. 113,994 113,994 113,994 106,584 113,566 82,874
R2 0.912 0.574 0.278 0.095 0.129 0.981Table 24.1: Results from various approaches to extreme values
| Wins, w/ROA | Raw | Wins | Trunc | Cook's D | Robust | |
|---|---|---|---|---|---|---|
| (Intercept) | -0.016*** | 0.324 | -0.030*** | -0.055*** | -0.058*** | 0.004*** |
| (-6.522) | (1.419) | (-3.400) | (-9.244) | (-6.881) | (3.145) | |
| big_nTRUE | 0.001 | -0.072 | 0.007 | 0.009** | 0.020*** | -0.010*** |
| (0.709) | (-1.067) | (1.444) | (2.758) | (4.023) | (-18.721) | |
| roa | 0.792*** | |||||
| (58.163) | ||||||
| cfo | -0.765*** | 0.801* | 0.260*** | 0.127*** | 0.194*** | -0.386*** |
| (-34.828) | (1.764) | (9.055) | (5.452) | (5.337) | (-173.511) | |
| size | 0.003*** | -0.065* | 0.004** | 0.006*** | 0.001 | 0.003*** |
| (5.800) | (-1.839) | (2.642) | (5.658) | (0.615) | (13.106) | |
| lev | -0.034** | -0.057 | -0.109*** | -0.030*** | -0.064*** | -0.020*** |
| (-2.636) | (-0.279) | (-5.030) | (-4.253) | (-4.453) | (-24.150) | |
| mtb | -0.002*** | -0.003 | -0.007*** | -0.006*** | -0.001*** | -0.000 |
| (-5.260) | (-0.927) | (-8.406) | (-10.867) | (-4.017) | (-0.675) | |
| Num.Obs. | 114196 | 114196 | 114196 | 107013 | 113766 | 83307 |
| R2 | 0.908 | 0.574 | 0.281 | 0.095 | 0.129 | 0.972 |
Cause: Code/reproducibility changes; full numerical split unresolved. The unadjusted OLS simulation now includes x2 and x3 as specified in the model, and simulation/robust-fit reproducibility was tightened. The robust row also depends on random draws and numerical fitting. External data are not used in this simulation. Source correction.
Effect on interpretation: Unchanged: all approaches over-reject in this setting, and robust regression performs best on test size but remains badly missized (40% at the nominal 5% level, compared with 52% in print).
Example: Robust-regression p_5 rejection rate: 0.52 in the hard-copy edition; 0.40 online.
Open current online Table 24.8
Partial replication of Panel B of Table 5 of Leone et al. (2019)
model est p_5 p_1
Do nothing 0.70 1.00 1.00
Winsorize all variables 0.70 1.00 1.00
Winsorize only X 0.72 1.00 1.00
Truncate all variables 0.68 1.00 1.00
Truncate only X 0.71 1.00 1.00
Cook’s distance 0.76 0.93 0.83
Robust regression 0.77 0.52 0.28Table 24.8: Partial replication of Panel B of Table 5 of Leone et al. ( 2019 )
| model | est | p_5 | p_1 |
|---|---|---|---|
| Do nothing | 0.70 | 1.00 | 1.00 |
| Winsorize all variables | 0.70 | 1.00 | 1.00 |
| Winsorize only X | 0.72 | 1.00 | 1.00 |
| Truncate all variables | 0.68 | 1.00 | 1.00 |
| Truncate only X | 0.71 | 1.00 | 1.00 |
| Cook’s distance | 0.76 | 0.92 | 0.82 |
| Robust regression | 0.78 | 0.40 | 0.16 |
Cause: Updated data. The auditor market-share sample remains fiscal 2019 with the same sample criteria. A historical Compustat refresh changes one Big Four count in bin (7,8].
Effect on interpretation: Unchanged: Big Four auditors dominate among large firms.
Example: Big 4 firms in market-cap bin (7,8]: 337 in the hard-copy edition; 338 online.
Open current online Table 25.2
Big 4 market share by market capitalization
bin big4_perc big4_num non_big4_num
(6,7] 16.32 63 323
(7,8] 27.53 337 887
(8,9] 58.03 1051 760
(9,10] 89.67 1580 182
(10,11] 96.76 687 23
(11,12] 97.85 91 2
(12,13] 100.00 1 0Table 25.2: Big 4 market share by market capitalization
| bin | big4_perc | big4_num | non_big4_num |
|---|---|---|---|
| (6,7] | 16.32 | 63 | 323 |
| (7,8] | 27.59 | 338 | 887 |
| (8,9] | 58.06 | 1052 | 760 |
| (9,10] | 89.68 | 1581 | 182 |
| (10,11] | 96.76 | 687 | 23 |
| (11,12] | 97.85 | 91 | 2 |
| (12,13] | 100.00 | 1 | 0 |
Cause: Updated empirical inputs and simulation ordering. The simulation samples the empirical market-capitalization distribution, which has changed. Deterministic row ordering also affects which observations are drawn under the same seed. This is not a pure external-data-free simulation.
Effect on interpretation: Unchanged: the positive Big Four coefficient is spurious because the simulated causal effect is zero and the linear market-capitalization control is misspecified.
Example: big4TRUE coefficient: 2.328*** in the hard-copy edition; 2.398*** online.
Open current online Table 25.4
Regression of audit quality using
simulated data
(1)
(Intercept) 1.567***
(0.052)
big4TRUE 2.328***
(0.066)
I(mkt_cap/1e+09) 0.053***
(0.001)
Num.Obs. 5,000
R2 0.542Table 25.4: Regression of audit quality using simulated data
| (1) | |
|---|---|
| (Intercept) | 1.542*** |
| (0.052) | |
| big4TRUE | 2.398*** |
| (0.066) | |
| I(mkt_cap/1e+09) | 0.053*** |
| (0.001) | |
| Num.Obs. | 5000 |
| R2 | 0.550 |
Cause: Caliper correction isolated; remaining input/order effects. On one fixed current simulated sample, the printed standardized 0.03 caliper yields N=2,298 and Big Four coefficients 0.233 and 0.218; the corrected probability-unit caliper yields N=2,626 and coefficients 0.523 and 0.501. Printed N=2,348, so the total difference also includes unresolved input/order effects. Source correction.
Effect on interpretation: Unchanged: matching still leaves a positive, statistically significant Big Four coefficient even though the simulated causal effect is zero. The matched examples and coefficient magnitudes change.
Example: ID of first illustrated matched observation: 1896 in the hard-copy edition; 798 online.
Open current online Table 25.5
Sample matched pairs
id big4 mkt_cap pscore subclass
1896 FALSE 3,474,299 0.527 220
2424 TRUE 11,966,370 0.526 220
2387 FALSE 877,821,520 0.444 333
2434 TRUE 877,805,370 0.444 333
2920 FALSE 216,822,120,000 0.000 481
2904 TRUE 384,003,796,090 0.000 481Table 25.5: Sample matched pairs
| id | big4 | mkt_cap | pscore | subclass |
|---|---|---|---|---|
| 798 | FALSE | 3,717,588,160 | 0.196 | 220 |
| 109 | TRUE | 3,709,944,000 | 0.196 | 220 |
| 1194 | FALSE | 67,900,230 | 0.539 | 333 |
| 1487 | TRUE | 347,568,000 | 0.509 | 333 |
| 1752 | FALSE | 7,596,000 | 0.546 | 481 |
| 3531 | TRUE | 63,895,520 | 0.540 | 481 |
Cause: Caliper correction isolated; remaining input/order effects. On one fixed current simulated sample, the printed standardized 0.03 caliper yields N=2,298 and Big Four coefficients 0.233 and 0.218; the corrected probability-unit caliper yields N=2,626 and coefficients 0.523 and 0.501. Printed N=2,348, so the total difference also includes unresolved input/order effects. Source correction.
Effect on interpretation: Unchanged: matching still leaves a positive, statistically significant Big Four coefficient even though the simulated causal effect is zero. The matched examples and coefficient magnitudes change.
Example: Observations in each matched regression: 2,348 in the hard-copy edition; 2,626 online.
Open current online Table 25.6
Relation between big4 and audit quality
(matched)
(1) (2)
(Intercept) 1.870*** 1.781***
(0.051) (0.045)
big4TRUE 0.235*** 0.220***
(0.072) (0.064)
I(mkt_cap/1e+09) 0.085***
(0.003)
Num.Obs. 2,348 2,348
R2 0.004 0.227Table 25.6: Relation between big4 and audit quality (matched)
| (1) | (2) | |
|---|---|---|
| (Intercept) | 1.648*** | 1.576*** |
| (0.047) | (0.042) | |
| big4TRUE | 0.523*** | 0.501*** |
| (0.066) | (0.059) | |
| I(mkt_cap/1e+09) | 0.084*** | |
| (0.003) | ||
| Num.Obs. | 2626 | 2626 |
| R2 | 0.023 | 0.218 |
Cause: Code/sample corrections plus input changes. Prior-year observations are retained through lag calculations; industry models require ten finite observations; and ties in ROA matching select one counterpart deterministically. Revised historical data and software can also contribute. Source correction.
Effect on interpretation: Material qualification: the untrimmed PSM Big Four estimate changes from -0.007*** to -0.005 without stars; the winsorized PSM estimate remains -0.006***. The chapter now explicitly says the significance conclusion depends on the specification.
Example: big4TRUE coefficient, PSM: −0.007*** in the hard-copy edition; −0.005 online.
Open current online Table 25.7
Relation between Big 4 and absolute value of discretionary accruals
OLS OLS, win PSM PSM, win
big4TRUE −0.016*** −0.014*** −0.007*** −0.006***
(0.002) (0.001) (0.002) (0.002)
log_mkt −0.006*** −0.005*** −0.007*** −0.006***
(0.001) (0.000) (0.001) (0.001)
lag_roa −0.108*** −0.124*** −0.111*** −0.135***
(0.006) (0.005) (0.017) (0.005)
lag_lev −0.045*** −0.045*** −0.055*** −0.052***
(0.003) (0.002) (0.005) (0.005)
lag_curr 0.000*** −0.001*** 0.000 −0.001*
(0.000) (0.000) (0.000) (0.000)
Num.Obs. 72,996 72,996 24,217 24,227
R2 0.127 0.150 0.117 0.132Table 25.7: Relation between Big 4 and absolute value of discretionary accruals
| OLS | OLS, win | PSM | PSM, win | |
|---|---|---|---|---|
| big4TRUE | -0.031*** | -0.017*** | -0.005 | -0.006*** |
| (0.002) | (0.002) | (0.004) | (0.002) | |
| log_mkt | -0.011*** | -0.006*** | -0.009*** | -0.006*** |
| (0.000) | (0.000) | (0.002) | (0.001) | |
| lag_roa | -0.002*** | -0.137*** | -0.066*** | -0.152*** |
| (0.000) | (0.005) | (0.015) | (0.006) | |
| lag_lev | -0.000 | -0.031*** | -0.006* | -0.026*** |
| (0.000) | (0.003) | (0.003) | (0.005) | |
| lag_curr | -0.000** | -0.001** | -0.000 | -0.000 |
| (0.000) | (0.000) | (0.000) | (0.000) | |
| Num.Obs. | 92064 | 92064 | 31635 | 31612 |
| R2 | 0.058 | 0.166 | 0.070 | 0.155 |
Cause: Code/package corrections plus input changes. The farr prediction method applied tree weights to the wrong observations through R vector recycling. Correcting the weighting invalidated saved boosting results and changed tuning choices. As-of fraud labels, test-label joins and cache dependencies were also corrected; refreshed features can contribute. This is not simply updated data. Source correction.
Effect on interpretation: Material change: mean test AUC now favours AdaBoost (0.696 versus 0.676), reversing the printed AUC ranking (0.623 versus 0.698). AdaBoost also has higher mean NDCG at 3%; RUSBoost has higher mean NDCG at 1% and 2%. The text already reflects this measure-dependent ranking.
Example: Best displayed AdaBoost validation AUC: 0.646 in the hard-copy edition; 0.649 online.
Open current online Table 26.1
AdaBoost cross-validation results for
different meta-parameters
size learn_rate maxdepth auc
30 0.100 5 0.646
30 0.100 7 0.644
50 0.100 7 0.642
50 0.100 5 0.624
50 0.100 3 0.624
50 1.000 5 0.611Table 26.1: AdaBoost cross-validation results for different meta-parameters
| size | learn_rate | maxdepth | auc |
|---|---|---|---|
| 30 | 1.000 | 3 | 0.649 |
| 50 | 0.100 | 7 | 0.643 |
| 50 | 1.000 | 3 | 0.641 |
| 50 | 0.100 | 3 | 0.638 |
| 30 | 0.500 | 5 | 0.631 |
| 30 | 0.100 | 5 | 0.629 |
Cause: Code/package corrections plus input changes. The farr prediction method applied tree weights to the wrong observations through R vector recycling. Correcting the weighting invalidated saved boosting results and changed tuning choices. As-of fraud labels, test-label joins and cache dependencies were also corrected; refreshed features can contribute. This is not simply updated data. Source correction.
Effect on interpretation: Material change: mean test AUC now favours AdaBoost (0.696 versus 0.676), reversing the printed AUC ranking (0.623 versus 0.698). AdaBoost also has higher mean NDCG at 3%; RUSBoost has higher mean NDCG at 1% and 2%. The text already reflects this measure-dependent ranking.
Example: Best displayed RUSBoost validation AUC: 0.694 in the hard-copy edition; 0.700 online.
Open current online Table 26.2
RUSBoost cross-validation results for different
meta-parameters
rus size learn_rate maxdepth ir auc
True 300 0.010 15 2 0.694
True 200 0.010 15 2 0.693
True 200 0.010 5 2 0.691
True 200 0.010 10 2 0.691
True 300 0.010 10 2 0.690
True 100 0.010 15 2 0.690Table 26.2: RUSBoost cross-validation results for different meta-parameters
| rus | size | learn_rate | maxdepth | ir | auc |
|---|---|---|---|---|---|
| TRUE | 100 | 0.500 | 5 | 1 | 0.700 |
| TRUE | 200 | 0.500 | 5 | 1 | 0.700 |
| TRUE | 300 | 0.500 | 5 | 1 | 0.697 |
| TRUE | 200 | 0.100 | 5 | 2 | 0.696 |
| TRUE | 300 | 0.100 | 5 | 2 | 0.689 |
| TRUE | 300 | 0.100 | 10 | 1 | 0.687 |
Cause: Code/package corrections plus input changes. The farr prediction method applied tree weights to the wrong observations through R vector recycling. Correcting the weighting invalidated saved boosting results and changed tuning choices. As-of fraud labels, test-label joins and cache dependencies were also corrected; refreshed features can contribute. This is not simply updated data. Source correction.
Effect on interpretation: Material change: mean test AUC now favours AdaBoost (0.696 versus 0.676), reversing the printed AUC ranking (0.623 versus 0.698). AdaBoost also has higher mean NDCG at 3%; RUSBoost has higher mean NDCG at 1% and 2%. The text already reflects this measure-dependent ranking.
Example: Mean AdaBoost test AUC: 0.623 in the hard-copy edition; 0.696 online.
Open current online Table 26.3
Results: AdaBoost
test_year auc ndcg_0.01 ndcg_0.02 ndcg_0.03
2003 0.666 0.027 0.063 0.063
2004 0.598 0.021 0.021 0.031
2005 0.678 0.000 0.000 0.023
2006 0.637 0.038 0.038 0.038
2007 0.570 0.000 0.000 0.000
2008 0.592 0.030 0.030 0.030
Mean 0.623 0.019 0.025 0.031Table 26.3: Results: AdaBoost
| test_year | auc | ndcg_0.01 | ndcg_0.02 | ndcg_0.03 |
|---|---|---|---|---|
| 2003 | 0.736 | 0.013 | 0.020 | 0.038 |
| 2004 | 0.726 | 0.027 | 0.049 | 0.069 |
| 2005 | 0.714 | 0.000 | 0.027 | 0.038 |
| 2006 | 0.717 | 0.021 | 0.053 | 0.053 |
| 2007 | 0.645 | 0.000 | 0.000 | 0.000 |
| 2008 | 0.638 | 0.030 | 0.030 | 0.046 |
| Mean | 0.696 | 0.015 | 0.030 | 0.041 |
Cause: Code/package corrections plus input changes. The farr prediction method applied tree weights to the wrong observations through R vector recycling. Correcting the weighting invalidated saved boosting results and changed tuning choices. As-of fraud labels, test-label joins and cache dependencies were also corrected; refreshed features can contribute. This is not simply updated data. Source correction.
Effect on interpretation: Material change: mean test AUC now favours AdaBoost (0.696 versus 0.676), reversing the printed AUC ranking (0.623 versus 0.698). AdaBoost also has higher mean NDCG at 3%; RUSBoost has higher mean NDCG at 1% and 2%. The text already reflects this measure-dependent ranking.
Example: Mean RUSBoost test AUC: 0.698 in the hard-copy edition; 0.676 online.
Open current online Table 26.4
Results: RUSBoost
test_year auc ndcg_0.01 ndcg_0.02 ndcg_0.03
2003 0.729 0.054 0.060 0.085
2004 0.728 0.029 0.039 0.059
2005 0.716 0.020 0.032 0.032
2006 0.728 0.000 0.015 0.015
2007 0.647 0.000 0.000 0.000
2008 0.639 0.023 0.023 0.023
Mean 0.698 0.021 0.028 0.036Table 26.4: Results: RUSBoost
| test_year | auc | ndcg_0.01 | ndcg_0.02 | ndcg_0.03 |
|---|---|---|---|---|
| 2003 | 0.726 | 0.031 | 0.047 | 0.056 |
| 2004 | 0.734 | 0.031 | 0.051 | 0.051 |
| 2005 | 0.648 | 0.024 | 0.050 | 0.062 |
| 2006 | 0.663 | 0.000 | 0.016 | 0.030 |
| 2007 | 0.671 | 0.000 | 0.000 | 0.000 |
| 2008 | 0.617 | 0.021 | 0.021 | 0.021 |
| Mean | 0.676 | 0.018 | 0.031 | 0.037 |
Cause: Label/join corrections plus input changes. The logit exercise shares corrected as-of training labels and a test-label join that gives one binary outcome per firm-year. Its features have also been refreshed. The boosting weight correction does not apply to logit. Source correction.
Effect on interpretation: Mean test AUC increases slightly (0.646 to 0.651) and remains below both boosted models. The exercise asks the reader to compare performance rather than asserting a contradictory ranking.
Example: Mean logit test AUC: 0.646 in the hard-copy edition; 0.651 online.
Open current online Table 26.5
Performance statistics for logit models
test_year auc ndcg_0.01 ndcg_0.02 ndcg_0.03
2003 0.668 0.000 0.029 0.046
2004 0.673 0.000 0.000 0.020
2005 0.654 0.000 0.013 0.024
2006 0.646 0.000 0.016 0.016
2007 0.598 0.000 0.000 0.000
2008 0.635 0.060 0.060 0.077
Mean 0.646 0.010 0.020 0.031Table 26.5: Performance statistics for logit models
| test_year | auc | ndcg_0.01 | ndcg_0.02 | ndcg_0.03 |
|---|---|---|---|---|
| 2003 | 0.679 | 0.012 | 0.030 | 0.030 |
| 2004 | 0.669 | 0.000 | 0.000 | 0.020 |
| 2005 | 0.670 | 0.000 | 0.013 | 0.024 |
| 2006 | 0.644 | 0.000 | 0.015 | 0.015 |
| 2007 | 0.597 | 0.000 | 0.000 | 0.000 |
| 2008 | 0.646 | 0.052 | 0.052 | 0.069 |
| Mean | 0.651 | 0.011 | 0.018 | 0.026 |
Initial analysis of student scores
treat pre post change
FALSE 515.539 519.657 4.118
TRUE 510.810 531.938 21.128Table 3.1: Initial analysis of student scores
| treat | pre | post | change |
|---|---|---|---|
| FALSE | 515.539 | 519.657 | 4.118 |
| TRUE | 510.810 | 531.938 | 21.128 |
Mean scores by grade
grade score
5 495.065
6 513.175
7 525.798
8 541.707Table 3.2: Mean scores by grade
| grade | score |
|---|---|
| 5 | 495.065 |
| 6 | 513.175 |
| 7 | 525.798 |
| 8 | 541.707 |
Test-score regressions with grade and id fixed effects
(1)
(Intercept) 495.370***
(2.527)
treatTRUE −2.235
(3.570)
postTRUE 38.774***
(0.276)
treatTRUE × postTRUE 15.736***
(0.319)
Num.Obs. 4,000Table 3.3: Test-score regressions with grade and id fixed effects
| (1) | |
|---|---|
| (Intercept) | 495.370*** |
| (2.527) | |
| treatTRUE | -2.235 |
| (3.570) | |
| postTRUE | 38.774*** |
| (0.276) | |
| treatTRUE × postTRUE | 15.736*** |
| (0.319) | |
| Num.Obs. | 4000 |
Regression with demeaned test scores
(1)
(Intercept) 3.934***
(0.138)
treatTRUE −7.868***
(0.195)
postTRUE −7.868***
(0.195)
treatTRUE × postTRUE 15.736***
(0.276)
Num.Obs. 4,000Table 3.4: Regression with demeaned test scores
| (1) | |
|---|---|
| (Intercept) | 3.934*** |
| (0.138) | |
| treatTRUE | -7.868*** |
| (0.195) | |
| postTRUE | -7.868*** |
| (0.195) | |
| treatTRUE × postTRUE | 15.736*** |
| (0.276) | |
| Num.Obs. | 4000 |
Regression results with grade fixed effects
(1)
(Intercept) 496.833***
(0.274)
treatTRUE −3.536***
(0.316)
postTRUE 38.774***
(0.388)
factor(grade)6 18.110***
(0.316)
factor(grade)7 −15.909***
(0.316)
treatTRUE × postTRUE 15.736***
(0.448)
Num.Obs. 4,000Table 3.5: Regression results with grade fixed effects
| (1) | |
|---|---|
| (Intercept) | 496.833*** |
| (0.274) | |
| treatTRUE | -3.536*** |
| (0.316) | |
| postTRUE | 38.774*** |
| (0.388) | |
| factor(grade)6 | 18.110*** |
| (0.316) | |
| factor(grade)7 | -15.909*** |
| (0.316) | |
| treatTRUE × postTRUE | 15.736*** |
| (0.448) | |
| Num.Obs. | 4000 |
Demeaned test scores by grade and treat
grade treat score
5.0000 False 3.3377
5.0000 True −3.3377
6.0000 False 4.5302
6.0000 True −4.5302
7.0000 False −3.9746
7.0000 True 3.9746
8.0000 False −3.8933
8.0000 True 3.8933Table 3.6: Demeaned test scores by grade and treat
| grade | treat | score |
|---|---|---|
| 5.0000 | FALSE | 3.3377 |
| 5.0000 | TRUE | -3.3377 |
| 6.0000 | FALSE | 4.5302 |
| 6.0000 | TRUE | -4.5302 |
| 7.0000 | FALSE | -3.9746 |
| 7.0000 | TRUE | 3.9746 |
| 8.0000 | FALSE | -3.8933 |
| 8.0000 | TRUE | 3.8933 |
Demonstration of FWL theorem
(1) (2)
(Intercept) −0.017 0.000
(0.028) (0.004)
big_nTRUE 0.022*
(0.011)
cfo 0.141*** 0.141***
(0.008) (0.008)
size 0.000
(0.002)
lev −0.066***
(0.013)
mtb 0.000***
(0.000)
Num.Obs. 8,850 8,850
Table 3.7: Demonstration of FWL theorem
| (1) | (2) | |
|---|---|---|
| (Intercept) | -0.017 | -0.000 |
| (0.028) | (0.004) | |
| big_nTRUE | 0.022* | |
| (0.011) | ||
| cfo | 0.141*** | 0.141*** |
| (0.008) | (0.008) | |
| size | -0.000 | |
| (0.002) | ||
| lev | -0.066*** | |
| (0.013) | ||
| mtb | -0.000*** | |
| (0.000) | ||
| Num.Obs. | 8850 | 8850 |
The industry and intelligence rows appear in a different order; their corresponding coefficients agree.
Income regressed on education and other variables
(1) (2) (3)
(Intercept) 9.997 10.003 10.019
education 6.767 4.999
intelligence 6.003 21.008
industry 7.000 27.007
Num.Obs. 100,000 100,000 100,000
R2 0.996 0.999 0.978Table 4.1: Income regressed on education and other variables
| (1) | (2) | (3) | |
|---|---|---|---|
| (Intercept) | 9.997 | 10.003 | 10.019 |
| education | 6.767 | 4.999 | |
| industry | 7.000 | 27.007 | |
| intelligence | 6.003 | 21.008 | |
| Num.Obs. | 100000 | 100000 | 100000 |
| R2 | 0.996 | 0.999 | 0.978 |
Regression with collider variables
(1) (2) (3) (4)
(Intercept) 0.001 −0.161*** 2.137*** −0.161***
test −0.008*** −0.174*** −0.719*** −0.174***
admittedTRUE 2.298***
test × admittedTRUE −0.545***
Num.Obs. 100,000 90,000 10,000 100,000
R2 0.000 0.031 0.515 0.227Table 4.2: Regression with collider variables
| (1) | (2) | (3) | (4) | |
|---|---|---|---|---|
| (Intercept) | 0.001 | -0.161*** | 2.137*** | -0.161*** |
| test | -0.008*** | -0.174*** | -0.719*** | -0.174*** |
| admittedTRUE | 2.298*** | |||
| test × admittedTRUE | -0.545*** | |||
| Num.Obs. | 100000 | 90000 | 10000 | 100000 |
| R2 | 0.000 | 0.031 | 0.515 | 0.227 |
Discrimination regressions
(1) (2)
(Intercept) 3.013*** 1.003***
femaleTRUE −5.005*** −0.996***
occupation 1.999***
Num.Obs. 100,000 100,000
R2 0.556 0.911Table 4.3: Discrimination regressions
| (1) | (2) | |
|---|---|---|
| (Intercept) | 3.013*** | 1.003*** |
| femaleTRUE | -5.005*** | -0.996*** |
| occupation | 1.999*** | |
| Num.Obs. | 100000 | 100000 |
| R2 | 0.556 | 0.911 |
Regression output calculated “by hand”
b se t p
0.007540 0.006440 1.171 0.242
0.997647 0.006321 157.829 0.000Table 5.2: Regression output calculated “by hand”
| b | se | t | p |
|---|---|---|---|
| 0.007540 | 0.006440 | 1.171 | 0.242 |
| 0.997647 | 0.006321 | 157.829 | 0.000 |
Regression output calculated using lm()
Estimate Std. Error t value Pr(>|t|)
(Intercept) 0.007540 0.006440 1.171 0.242
x 0.997647 0.006321 157.829 0.000Table 5.3: Regression output calculated using lm()
| Estimate | Std. Error | t value | Pr(>|t|) | |
|---|---|---|---|---|
| (Intercept) | 0.007540 | 0.006440 | 1.171 | 0.242 |
| x | 0.997647 | 0.006321 | 157.829 | 0.000 |
Bootstrapped standard errors
(Intercept) x
0.0064542 0.0064964Table 5.4: Bootstrapped standard errors
| (Intercept) | x |
|---|---|
| 0.0064542 | 0.0064964 |
Step 1 of Fama-MacBeth: Coefficients by year
year (Intercept) x
1 1.6356 0.9514
2 2.7017 0.9899
3 −1.1093 0.9078
4 −0.3491 1.0952
5 −0.8629 1.0035
6 −0.4451 0.9104
7 −0.2011 1.0507
8 1.5738 0.8011
9 −2.7727 0.9517
10 0.8072 0.9843Table 5.5: Step 1 of Fama-MacBeth: Coefficients by year
| year | (Intercept) | x |
|---|---|---|
| 1 | 1.6356 | 0.9514 |
| 2 | 2.7017 | 0.9899 |
| 3 | -1.1093 | 0.9078 |
| 4 | -0.3491 | 1.0952 |
| 5 | -0.8629 | 1.0035 |
| 6 | -0.4451 | 0.9104 |
| 7 | -0.2011 | 1.0507 |
| 8 | 1.5738 | 0.8011 |
| 9 | -2.7727 | 0.9517 |
| 10 | 0.8072 | 0.9843 |
Standard errors using FM-t and FM-NW
FM-t FM-NW
(Intercept) x (Intercept) x
Estimate 0.098 0.965*** 0.098 0.965***
(0.506) (0.026) (0.457) (0.020)
Num.Obs. 10 10 10 10
R2 0.000 0.000 0.000 0.000Table 5.6: Standard errors using FM-t and FM-NW
| FM-t | FM-NW | |||
|---|---|---|---|---|
| (Intercept) | x | (Intercept) | x | |
| Estimate | 0.098 | 0.965*** | 0.098 | 0.965*** |
| (0.506) | (0.026) | (0.457) | (0.020) | |
| Num.Obs. | 10 | 10 | 10 | 10 |
| R2 | 0.000 | 0.000 | 0.000 | 0.000 |
FM-t using pmg()
(1)
(Intercept) 0.098
(0.506)
x 0.965***
(0.026)
Num.Obs. 5,000
R2 0.786Table 5.7: FM-t using pmg()
| (1) | |
|---|---|
| (Intercept) | 0.098 |
| (0.506) | |
| x | 0.965*** |
| (0.026) | |
| Num.Obs. | 5000 |
| R2 | 0.786 |
Standard errors using various methods
OLS White NW FM-t CL-i CL-t CL-2
(Intercept) 0.021 0.021 0.021 0.098 0.021 0.021 0.021
(0.026) (0.026) (0.027) (0.506) (0.023) (0.492) (0.492)
x 1.266*** 1.266*** 1.266*** 0.965*** 1.266*** 1.266*** 1.266***
(0.027) (0.021) (0.021) (0.026) (0.018) (0.172) (0.172)
Num.Obs. 5,000 5,000 5,000 5,000 5,000 5,000 5,000
R2 0.305 0.305 0.305 0.786 0.305 0.305 0.305Table 5.8: Standard errors using various methods
| OLS | White | NW | FM-t | CL-i | CL-t | CL-2 | |
|---|---|---|---|---|---|---|---|
| (Intercept) | 0.021 | 0.021 | 0.021 | 0.098 | 0.021 | 0.021 | 0.021 |
| (0.026) | (0.026) | (0.027) | (0.506) | (0.023) | (0.492) | (0.492) | |
| x | 1.266*** | 1.266*** | 1.266*** | 0.965*** | 1.266*** | 1.266*** | 1.266*** |
| (0.027) | (0.021) | (0.021) | (0.026) | (0.018) | (0.172) | (0.172) | |
| Num.Obs. | 5000 | 5000 | 5000 | 5000 | 5000 | 5000 | 5000 |
| R2 | 0.305 | 0.305 | 0.305 | 0.786 | 0.305 | 0.305 | 0.305 |
A sample of observations on crsp.ccmxpf_lnkhist gvkey linkprim liid linktype lpermno lpermco linkdt linkenddt 001000 C 00X NU NA NA 1961-01-01 1970-09-29 001000 P 01 NU NA NA 1970-09-30 1970-11-12 001000 P 01 LU 25881 23369 1970-11-13 1978-06-30 001001 C 00X NU NA NA 1978-01-01 1983-09-19 001001 P 01 LU 10015 6398 1983-09-20 1986-07-31 001002 C 00X NR NA NA 1960-01-01 1970-09-29 001002 C 01 NR NA NA 1970-09-30 1972-12-13 001002 C 01 NR NA NA 1973-06-06 1973-08-31 001002 C 01 LC 10023 22159 1972-12-14 1973-06-05 001003 C 00X NU NA NA 1980-01-01 1983-12-06
Table 7.2: A sample of observations on crsp.ccmxpf_lnkhist
| gvkey | linkprim | liid | linktype | lpermno | lpermco | linkdt | linkenddt |
|---|---|---|---|---|---|---|---|
| 001000 | C | 00X | NU | NA | NA | 1961-01-01 | 1970-09-29 |
| 001000 | P | 01 | NU | NA | NA | 1970-09-30 | 1970-11-12 |
| 001000 | P | 01 | LU | 25881 | 23369 | 1970-11-13 | 1978-06-30 |
| 001001 | C | 00X | NU | NA | NA | 1978-01-01 | 1983-09-19 |
| 001001 | P | 01 | LU | 10015 | 6398 | 1983-09-20 | 1986-07-31 |
| 001002 | C | 00X | NR | NA | NA | 1960-01-01 | 1970-09-29 |
| 001002 | C | 01 | NR | NA | NA | 1970-09-30 | 1972-12-13 |
| 001002 | C | 01 | NR | NA | NA | 1973-06-06 | 1973-08-31 |
| 001002 | C | 01 | LC | 10023 | 22159 | 1972-12-14 | 1973-06-05 |
| 001003 | C | 00X | NU | NA | NA | 1980-01-01 | 1983-12-06 |
Example of GVKEY mapping to multiple PERMNOs
gvkey linkprim liid linktype lpermno lpermco linkdt linkenddt
011550 P 01 LC 23536 21931 1962-01-31 NA
013353 P 01 LD 23536 21931 1962-01-31 1986-12-31
013353 C 99X LD 23536 21931 1987-01-01 2020-12-31Table 7.4: Example of GVKEY mapping to multiple PERMNOs
| gvkey | linkprim | liid | linktype | lpermno | lpermco | linkdt | linkenddt |
|---|---|---|---|---|---|---|---|
| 011550 | P | 01 | LC | 23536 | 21931 | 1962-01-31 | NA |
| 013353 | P | 01 | LD | 23536 | 21931 | 1962-01-31 | 1986-12-31 |
| 013353 | C | 99X | LD | 23536 | 21931 | 1987-01-01 | 2020-12-31 |
Example where linktype is LX gvkey linkprim liid linktype lpermno lpermco linkdt linkenddt 001186 P 01 LC 78223 26174 1982-11-01 NA 001186 N 01C LX 78223 26174 1982-11-01 NA
Table 7.5: Example where linktype is LX
| gvkey | linkprim | liid | linktype | lpermno | lpermco | linkdt | linkenddt |
|---|---|---|---|---|---|---|---|
| 001186 | P | 01 | LC | 78223 | 26174 | 1982-11-01 | NA |
| 001186 | N | 01C | LX | 78223 | 26174 | 1982-11-01 | NA |
Open current online Table 15.1
Translation of key pre-2006 Compustat items
Old item Current item Item description
#1 che Cash and Short-Term Investments
#4 act Current Assets—Total
#5 lct Current Liabilities—Total
#14 dp Depreciation and Amortization
#34 dlc Debt in Current Liabilities—TotalTable 15.1: Translation of key pre-2006 Compustat items
| Old item | Current item | Item description |
|---|---|---|
| #1 | che | Cash and Short-Term Investments |
| #4 | act | Current Assets—Total |
| #5 | lct | Current Liabilities—Total |
| #14 | dp | Depreciation and Amortization |
| #34 | dlc | Debt in Current Liabilities—Total |
Open current online Table 17.1
Number of observations for
top 5 natural disasters
eventdate n
2005-08-29 131
2008-09-13 54
2004-09-16 23
2004-08-13 11
2005-10-24 9Table 17.1: Number of observations for top 5 natural disasters
| eventdate | n |
|---|---|
| 2005-08-29 | 131 |
| 2008-09-13 | 54 |
| 2004-09-16 | 23 |
| 2004-08-13 | 11 |
| 2005-10-24 | 9 |
Open current online Table 19.1
Reconciliation of sample with Russell 3000
nasdaq_small recent_listing delisted keep n
FALSE FALSE FALSE TRUE 2954
FALSE FALSE TRUE FALSE 2
FALSE TRUE FALSE FALSE 12
TRUE FALSE FALSE FALSE 32Table 19.1: Reconciliation of sample with Russell 3000
| nasdaq_small | recent_listing | delisted | keep | n |
|---|---|---|---|---|
| FALSE | FALSE | FALSE | TRUE | 2954 |
| FALSE | FALSE | TRUE | FALSE | 2 |
| FALSE | TRUE | FALSE | FALSE | 12 |
| TRUE | FALSE | FALSE | FALSE | 32 |
Open current online Table 21.1
Regression analysis: Simulated data
(1) (2)
I(treat * post) 0.736** 0.736
(0.319) (0.995)
Num.Obs. 4,000 4,000
R2 0.923 0.923Table 21.1: Regression analysis: Simulated data
| (1) | (2) | |
|---|---|---|
| I(treat * post) | 0.736** | 0.736 |
| (0.319) | (0.995) | |
| Num.Obs. | 4000 | 4000 |
| R2 | 0.923 | 0.923 |
Open current online Table 21.2
Largest disclosed customers
gvkey datadate cnms salecs
002751 2010-06-30 CVS Health Corp 23,641
002751 2009-06-30 WALGREEN CO 22,888
118122 1999-12-31 General Motors Corp 22,302
002751 2010-06-30 WALGREEN CO 21,671
002751 2009-06-30 CVS Caremark Corp 20,898
118122 2000-12-31 General Motors Corp 20,665
002751 2008-06-30 CVS Caremark Corp 20,040
002751 2007-06-30 CVS Caremark Corp 18,239
012136 2005-06-30 GENERAL ELECTRIC CO 17,712
002751 2008-06-30 Walgreen Co 17,307Table 21.2: Largest disclosed customers
| gvkey | datadate | cnms | salecs |
|---|---|---|---|
| 002751 | 2010-06-30 | CVS Health Corp | 23,641 |
| 002751 | 2009-06-30 | WALGREEN CO | 22,888 |
| 118122 | 1999-12-31 | General Motors Corp | 22,302 |
| 002751 | 2010-06-30 | WALGREEN CO | 21,671 |
| 002751 | 2009-06-30 | CVS Caremark Corp | 20,898 |
| 118122 | 2000-12-31 | General Motors Corp | 20,665 |
| 002751 | 2008-06-30 | CVS Caremark Corp | 20,040 |
| 002751 | 2007-06-30 | CVS Caremark Corp | 18,239 |
| 012136 | 2005-06-30 | GENERAL ELECTRIC CO | 17,712 |
| 002751 | 2008-06-30 | Walgreen Co | 17,307 |
Open current online Table 21.3
Means of key disclosure measures
measure mean median sd
ratio 0.459 0.333 0.449
ratio_sale 0.461 0.316 0.465Table 21.3: Means of key disclosure measures
| measure | mean | median | sd |
|---|---|---|---|
| ratio | 0.459 | 0.333 | 0.449 |
| ratio_sale | 0.461 | 0.316 | 0.465 |
Open current online Table 22.1
Sample of firm-years appearing to meet criteria,
but no SOX 404 report
cik fdate pfdate publicfloat
1003472 2004-12-31 2005-03-09 141.9853
943861 2004-12-31 2005-03-24 168.5346
906780 2004-12-31 2005-03-01 171.6621
1085869 2004-12-31 2005-02-28 172.0916
29834 2004-12-31 2005-03-16 184.9020
1028205 2005-06-30 2005-08-15 279.1016Table 22.1: Sample of firm-years appearing to meet criteria, but no SOX 404 report
| cik | fdate | pfdate | publicfloat |
|---|---|---|---|
| 1003472 | 2004-12-31 | 2005-03-09 | 141.9853 |
| 943861 | 2004-12-31 | 2005-03-24 | 168.5346 |
| 906780 | 2004-12-31 | 2005-03-01 | 171.6621 |
| 1085869 | 2004-12-31 | 2005-02-28 | 172.0916 |
| 29834 | 2004-12-31 | 2005-03-16 | 184.9020 |
| 1028205 | 2005-06-30 | 2005-08-15 | 279.1016 |
Open current online Table 23.1
Relationships between model coefficients
Logit OLS mfx_logit mfx_OLS
(Intercept) 0.034*** 0.001*** 0.000 0.000
(9.937) (2.868) (−0.191) (−0.653)
coef_probit 2.172*** 0.057***
(202.626) (54.862)
mfx_probit 0.997*** 0.998***
(309.056) (235.526)
Num.Obs. 1,000 1,000 1,000 1,000
R2 0.976 0.751 0.990 0.982
Table 23.1: Relationships between model coefficients
| Logit | OLS | mfx_logit | mfx_OLS | |
|---|---|---|---|---|
| (Intercept) | 0.034*** | 0.001*** | -0.000 | -0.000 |
| (9.937) | (2.868) | (-0.191) | (-0.653) | |
| coef_probit | 2.172*** | 0.057*** | ||
| (202.626) | (54.862) | |||
| mfx_probit | 0.997*** | 0.998*** | ||
| (309.056) | (235.526) | |||
| Num.Obs. | 1000 | 1000 | 1000 | 1000 |
| R2 | 0.976 | 0.751 | 0.990 | 0.982 |
Open current online Table 23.2
Probit simulation results
type beta_1_true mean_coef mean_mfx rej_rate_5 rej_rate_1
OLS 0.0 0.000 0.000 0.035 0.008
logit 0.0 0.006 0.000 0.033 0.008
probit 0.0 0.002 0.000 0.035 0.008
OLS 0.3 0.018 0.018 0.938 0.805
logit 0.3 0.696 0.018 0.938 0.800
probit 0.3 0.305 0.018 0.934 0.801
Table 23.2: Probit simulation results
| type | beta_1_true | mean_coef | mean_mfx | rej_rate_5 | rej_rate_1 |
|---|---|---|---|---|---|
| OLS | 0.0 | 0.000 | 0.000 | 0.035 | 0.008 |
| logit | 0.0 | 0.006 | 0.000 | 0.033 | 0.008 |
| probit | 0.0 | 0.002 | 0.000 | 0.035 | 0.008 |
| OLS | 0.3 | 0.018 | 0.018 | 0.938 | 0.805 |
| logit | 0.3 | 0.696 | 0.018 | 0.938 | 0.800 |
| probit | 0.3 | 0.305 | 0.018 | 0.934 | 0.801 |
Open current online Table 23.3
Poisson simulation results
type beta_1_true mean_coef mean_mfx rej_rate_5 rej_rate_1
OLS 0.0 0.000 0.000 0.059 0.011
Poisson 0.0 0.001 0.000 0.063 0.010
OLS 0.3 0.043 0.043 0.942 0.849
Poisson 0.3 0.302 0.043 0.940 0.850Table 23.3: Poisson simulation results
| type | beta_1_true | mean_coef | mean_mfx | rej_rate_5 | rej_rate_1 |
|---|---|---|---|---|---|
| OLS | 0.0 | 0.000 | 0.000 | 0.059 | 0.011 |
| Poisson | 0.0 | 0.001 | 0.000 | 0.063 | 0.010 |
| OLS | 0.3 | 0.043 | 0.043 | 0.942 | 0.849 |
| Poisson | 0.3 | 0.302 | 0.043 | 0.940 | 0.850 |
Open current online Table 24.2
Top 1% observations as percentage of total penalties
target total total_top top_perc
emppenalty 21,176 15,877 75.0
empprisonmos 17,181 3,658 21.3
firmpenalty 13,578 7,396 54.5
otherpenalty 11,107 9,684 87.2Table 24.2: Top 1% observations as percentage of total penalties
| target | total | total_top | top_perc |
|---|---|---|---|
| emppenalty | 21,176 | 15,877 | 75.0 |
| empprisonmos | 17,181 | 3,658 | 21.3 |
| firmpenalty | 13,578 | 7,396 | 54.5 |
| otherpenalty | 11,107 | 9,684 | 87.2 |
Open current online Table 24.3
Enforcement outcomes (Table 4 of CMSW)
firmpenalty emppenalty empprisonmos
wbflag 1.250* 1.111*** 0.621**
1.949 2.637 2.372
Num.Obs. 658 658 658Table 24.3: Enforcement outcomes (Table 4 of CMSW)
| firmpenalty | emppenalty | empprisonmos | |
|---|---|---|---|
| wbflag | 1.250* | 1.111*** | 0.621** |
| 1.949 | 2.637 | 2.372 | |
| Num.Obs. | 658 | 658 | 658 |
Open current online Table 24.4
Enforcement outcomes with OLS (Table 8 of CMSW)
firmpenalty emppenalty empprisonmos
wbflag 0.484*** 0.072 0.181
3.316 0.627 1.118
Num.Obs. 658 658 658Table 24.4: Enforcement outcomes with OLS (Table 8 of CMSW)
| firmpenalty | emppenalty | empprisonmos | |
|---|---|---|---|
| wbflag | 0.484*** | 0.072 | 0.181 |
| 3.316 | 0.627 | 1.118 | |
| Num.Obs. | 658 | 658 | 658 |
Open current online Table 24.5
Impact threshold for a confound-
ing variable (ITCV)
alpha crit_val itcv
0.01 2.584 0.032
0.05 1.964 0.058
0.10 1.647 0.070Table 24.5: Impact threshold for a confounding variable (ITCV)
| alpha | crit_val | itcv |
|---|---|---|
| 0.01 | 2.584 | 0.032 |
| 0.05 | 1.964 | 0.058 |
| 0.10 | 1.647 | 0.070 |
Open current online Table 24.6
Impacts for control variables
var r_yz r_xz impact
lnmktcap 0.332 0.274 0.091
bribeflag 0.289 0.035 0.010
lnvioperiod 0.140 0.053 0.007
lev 0.070 0.100 0.007
blckownpct −0.092 −0.075 0.007
lnuscodecnt 0.158 0.036 0.006
mkt2bk −0.060 −0.079 0.005
misledflag −0.051 −0.079 0.004
initabret 0.098 0.040 0.004
recidivist 0.034 0.080 0.003Table 24.6: Impacts for control variables
| var | r_yz | r_xz | impact |
|---|---|---|---|
| lnmktcap | 0.332 | 0.274 | 0.091 |
| bribeflag | 0.289 | 0.035 | 0.010 |
| lnvioperiod | 0.140 | 0.053 | 0.007 |
| lev | 0.070 | 0.100 | 0.007 |
| blckownpct | −0.092 | −0.075 | 0.007 |
| lnuscodecnt | 0.158 | 0.036 | 0.006 |
| mkt2bk | −0.060 | −0.079 | 0.005 |
| misledflag | −0.051 | −0.079 | 0.004 |
| initabret | 0.098 | 0.040 | 0.004 |
| recidivist | 0.034 | 0.080 | 0.003 |
Open current online Table 24.7
Combined impacts of control variables
var r_yz r_xz impact
z 0.553 0.494 0.273Table 24.7: Combined impacts of control variables
| var | r_yz | r_xz | impact |
|---|---|---|---|
| z | 0.553 | 0.494 | 0.273 |
Open current online Table 25.1
Values of aucd for top auditors
aucd audesc
0 Unaudited
1 Arthur Andersen
2 Arthur Young
3 Coopers & Lybrand
4 Ernst & Young
5 Deloitte & Touche
6 KPMG Peat Marwick
7 PricewaterhouseCoopers
8 Touche Ross
9 Other
10 Altschuler, Melvoin, and Glasser
11 BDO SeidmanTable 25.1: Values of aucd for top auditors
| aucd | audesc |
|---|---|
| 0 | Unaudited |
| 1 | Arthur Andersen |
| 2 | Arthur Young |
| 3 | Coopers & Lybrand |
| 4 | Ernst & Young |
| 5 | Deloitte & Touche |
| 6 | KPMG Peat Marwick |
| 7 | PricewaterhouseCoopers |
| 8 | Touche Ross |
| 9 | Other |
| 10 | Altschuler, Melvoin, and Glasser |
| 11 | BDO Seidman |
Open current online Table 25.3
Two large non-Big 4 firms gvkey datadate fyear au prcc_f csho big4 mkt_cap 014447 2019-12-31 2019 9 92.988 2,518.262 FALSE 234,168,146,856 026478 2019-12-31 2019 9 14.380 8,550.000 FALSE 122,949,000,000
Table 25.3: Two large non-Big 4 firms
| gvkey | datadate | fyear | au | prcc_f | csho | big4 | mkt_cap |
|---|---|---|---|---|---|---|---|
| 014447 | 2019-12-31 | 2019 | 9 | 92.988 | 2,518.262 | FALSE | 234,168,146,856 |
| 026478 | 2019-12-31 | 2019 | 9 | 14.380 | 8,550.000 | FALSE | 122,949,000,000 |
Variables in the Australian bank data
Item Description
gvkey Firm identifier
ticker Stock exchange ticker
co_name Bank’s name
ret Stock return (monthly)
mkt_cap Market capitalization
ni Net income
ib Income before extraordinary items
xi Extraordinary items
do Discontinued itemsTable 2.1: Variables in the Australian bank data
| Item | Description |
|---|---|
| gvkey | Firm identifier |
| ticker | Stock exchange ticker |
| co_name | Bank’s name |
| ret | Stock return (monthly) |
| mkt_cap | Market capitalization |
| ni | Net income |
| ib | Income before extraordinary items |
| xi | Extraordinary items |
| do | Discontinued items |
Firm and security identifiers
Data provider Firm identifiers Notes
Compustat (comp) gvkey
CRSP (crsp) permno, permco permno is a security identifier
IBES (ibes) ticker ticker is not necessarily the
ticker assigned to the firm by the
exchange on which it trades
SEC EDGAR CIK
Audit Analytics (audit) company_fkey company_fkey is the same as CIK
Various CUSIP CUSIP is a security identifierTable 7.1: Firm and security identifiers
| Data provider | Firm identifiers | Notes |
|---|---|---|
| Compustat ( comp ) | gvkey | |
| CRSP ( crsp ) | permno , permco | permno is a security identifier |
| IBES ( ibes ) | ticker | ticker is not necessarily the ticker assigned to the firm by the exchange on which it trades |
| SEC EDGAR | CIK | |
| Audit Analytics ( audit ) | company_fkey | company_fkey is the same as CIK |
| Various | CUSIP | CUSIP is a security identifier |
Components of total assets
Item Description
act Current Assets—Total
ppent Property Plant and Equipment—Total (Net)
ivaeq Investment and Advances—Equity
ivao Investment and Advances—Other
intan Intangible Assets—Total
ao Assets Other—Total
dc Deferred Charges (component of AO)
aox Assets Other (Sundry) (component of AO)Table 8.1: Components of total assets
| Item | Description |
|---|---|
| act | Current Assets—Total |
| ppent | Property Plant and Equipment—Total (Net) |
| ivaeq | Investment and Advances—Equity |
| ivao | Investment and Advances—Other |
| intan | Intangible Assets—Total |
| ao | Assets Other—Total |
| dc | Deferred Charges (component of AO) |
| aox | Assets Other (Sundry) (component of AO) |
Components of cash balance
Item Description
ch Cash
ivst Short-Term Investments
che Cash and Short-Term Investments—TotalTable 8.2: Components of cash balance
| Item | Description |
|---|---|
| ch | Cash |
| ivst | Short-Term Investments |
| che | Cash and Short-Term Investments—Total |
Components of cash flow statement
Item Description
oancf Operating Activities—Net Cash Flow
ivncf Investing Activities—Net Cash Flow
fincf Financing Activities—Net Cash Flow
exre Exchange Rate Effect
chech Cash and Cash Equivalents—Increase (Decrease)Table 8.3: Components of cash flow statement
| Item | Description |
|---|---|
| oancf | Operating Activities—Net Cash Flow |
| ivncf | Investing Activities—Net Cash Flow |
| fincf | Financing Activities—Net Cash Flow |
| exre | Exchange Rate Effect |
| chech | Cash and Cash Equivalents—Increase (Decrease) |
Open current online Table 14.1
Date variables on comp.funda
Variable Description
datadate Period-end date
apdedate Actual period-end date
pdate Date the data are updated on a preliminary basis
fdate Date the data are finalizedTable 14.1: Date variables on comp.funda
| Variable | Description |
|---|---|
| datadate | Period-end date |
| apdedate | Actual period-end date |
| pdate | Date the data are updated on a preliminary basis |
| fdate | Date the data are finalized |
SQL translations of key dplyr verbs
dplyr verb (R) SQL equivalent
|> FROM
select() SELECT
filter() WHERE
group_by() GROUP BY
arrange() ORDER BY
mutate() used-defined columns
summarize() used-defined aggregate columnsTable B.1: SQL translations of key dplyr verbs
| dplyr verb (R) | SQL equivalent |
|---|---|
| |> | FROM |
| select() | SELECT |
| filter() | WHERE |
| group_by() | GROUP BY |
| arrange() | ORDER BY |
| mutate() | user-defined columns |
| summarize() | user-defined aggregate columns |