SEMINAR FINAL EXAM

Ghulam Mustafa 27297

 Q2
We here are doing correlation and regression analysis
Fama-MacBeth (1973) Two-Step procedure
AR Coef. St.Err. t-value p-value [95% Conf Interval] Sig
LTA -.798 .721 -1.11 .292 -2.385 .789
LOP .673 .641 1.05 .316 -.738 2.083
Constant 3.875 2.726 1.42 .183 -2.125 9.874

Mean dependent var 0.300 SD dependent var 0.747

R-squared 0.150 Number of obs 922

F-test 3.875 Prob > F 0.053

*** p<.01, ** p<.05, * p<.1

• LTA: The coefficient for LTA is -0.798, indicating a positive relationship with the dependent
variable. However, this result is not statistically significant (p-value = 0.292), suggesting insufficient
evidence to support a significant relationship between LTA and the dependent variable.

• LOP: The coefficient for LOP 0.673, suggesting a positive relationship with the dependent
variable. However, this result is not statistically significant (p-value = 0.316), indicating insufficient
evidence to support a significant relationship between LOP and the dependent variable.

• Constant: The constant term has a coefficient of 3.875, indicating a positive intercept. It is
not statistically significant (p-value = 0.183), with a confidence interval of 95%

The F- test have a value of 3.875 and P-Value of 5.3% which is greater then 5% meaning that model
is not the best fit.

  OP Returns  
Date OP1 OP2 OP3 OP4 OP5
31/12/20 2.08653 0.97855 0.83238 0.42698 -
12 5 6 1 5 1.65955
31/12/20 0.75819 0.58643 0.57176 -
13 7 1.22761 8 5 0.18643
31/12/20 0.65141 0.56448 0.70702 0.23496 -
14 1 7 8 5 0.41645
31/12/20 0.37799 0.04711 0.12767 -
15 1 7 7 0.0287 0.34929
30/12/20 0.54269 0.83210 0.62338 0.38886 -
16 3 6 3 7 0.15383
29/12/20 0.62978 - - -
17 6 -0.1479 0.19199 0.06078 0.69057
 31/12/20 0.05928 - - 0.00859 -
18 7 0.10385 0.06137 3 0.05069
31/12/20 0.04411 - 0.00847 0.07838 0.03427
19 1 0.02627 6 8 7
31/12/20 0.54890 0.63745 0.36258 - -
20 4 7 3 0.00535 0.55425
31/12/20 0.07394 - 0.06880 0.07494 0.00100
21 1 0.02151 6 2 1
0.57728 0.30634 0.17470 -
Average 6 0.39878 1 8 0.40258

The average of highest extreme values is shown in OP4 and OP1. We found that at extreme lowest
level, we took 30Th percentile and the average of the values are 0.57. At the highest percentile of 70 Th
percentile, we saw the average value of 0.174. The difference between the OP4 and OP1 is -
0.402 which means that the returns at lowest percentile are way higher than the above percentile
value. So the premium is negative if we have to observe the returns of OP4 over OP1 and there
seems to be no premium associated with it.

Total Asset Return  

Date TA1 TA2 TA3 TA4 TA5
31/12/20 1.86140 0.63531 1.19383 0.63973 -
12 3 5 2 5 1.22167
31/12/20 1.09589 0.60290 0.49354 -
13 0.9268 5 5 3 0.43326
31/12/20 0.63356 0.77100 0.43898 0.28670 -
14 5 3 3 3 0.34686
31/12/20 0.26430 0.16079 - -
15 7 3 0.2483 0.01636 0.28066
30/12/20 0.48854 0.66502 0.76090 0.46615 -
16 2 4 1 1 0.02239
29/12/20 0.61565 - - - -
17 1 0.05274 0.16403 0.12092 0.73657
31/12/20 0.03997 - 0.05755 0.01758
18 4 -0.0778 0.13612 9 5
31/12/20 0.05737 - - 0.02255
19 5 0.00138 0.03848 0.07993 5
31/12/20 0.49849 0.64198 0.36396 0.04131 -
20 6 1 9 6 0.45718
31/12/20 0.11622 - 0.02315 0.06520 -
21 1 0.02724 9 2 0.05102
0.55023 0.38108 0.32934 0.19928 -
Average 3 5 1 6 0.35095

The average of highest extreme values is shown in TA4 and TA1. We found that at extreme lowest
level, we took 30Th percentile and the average of the values are 0.55. At the highest percentile of 70 Th
percentile, we saw the average value of 0.199. The difference between the TA4 and TA1 is -
0.350 which means that the returns at lowest percentile are way higher than the above percentile
value. So, the premium is negative if we must observe the returns of TA4 and TA1 and there seems
to be no premium associated with it.

Q3
Code Stata:
gen Month = month(Date)
sort ID Date
bys ID: gen Time =_n
xtset ID Time
gen LMV = ln(MV)
gen LVO = ln(VO)
gen PTBV2 =(PTBV/100)
asdoc pwcorr MR LMV PTBV2 PE EPS LVO , save(Quest3Ghulam.doc)
asdoc xtfmb MR LMV PTBV2 PE EPS LVO , save(Quest3xtfmbGhulam.doc)

Code Panel:
Sub Panel2()

Dim lr As Integer
Dim c As Integer
Dim g As Integer
Dim ID As Integer
Dim x As Variant
Dim D As Date

Application.ScreenUpdating = False
lr = Sheets("MR").Cells(Rows.Count, 1).End(xlUp).Row
lc = Sheets("MR").Cells(1, Columns.Count).End(xlToLeft).Column
Sheets.Add after:=Sheets(Sheets.Count)
Sheets(ActiveSheet.Name).Name = "Panel2"
x=2
For g = 2 To lr ' essential to take each row into the columns
ID = 1 'firms MRe given unique number as its position in sample
For c = 2 To lc ' Take all necessMRy information across firms
D = Sheets("MR").Cells(g, 1).Value
ret = Sheets("MR").Cells(g, c).Value
MV = Sheets("MV").Cells(g, c).Value
PTBV = Sheets("PTBV").Cells(g, c).Value
PE = Sheets("PE").Cells(g, c).Value
EPS = Sheets("EPS").Cells(g, c).Value
VO = Sheets("VO").Cells(g, c).Value
If MV <> "" And PTBV <> "" And PE <> "" And EPS <> "" And VO <> "" Then
Sheets("Panel2").Cells(x, 1).Value = D
Sheets("Panel2").Cells(x, 2).Value = ID
Sheets("Panel2").Cells(x, 3).Value = ret
Sheets("Panel2").Cells(x, 4).Value = MV
Sheets("Panel2").Cells(x, 5).Value = PTBV
 Sheets("Panel2").Cells(x, 6).Value = PE
Sheets("Panel2").Cells(x, 7).Value = EPS
Sheets("Panel2").Cells(x, 8).Value = VO

x = x + 1 ' It will when all columns MRe exhausted

End If

ID = ID + 1
Next c
Next g 'repeat each month the same thing
Sheets("Panel2").Cells(1, 1).Value = "Date"
Sheets("Panel2").Cells(1, 2).Value = "ID"
Sheets("Panel2").Cells(1, 3).Value = "MR"
Sheets("Panel2").Cells(1, 4).Value = "MV"
Sheets("Panel2").Cells(1, 5).Value = "PTBV"
Sheets("Panel2").Cells(1, 6).Value = "PE"
Sheets("Panel2").Cells(1, 7).Value = "EPS"
Sheets("Panel2").Cells(1, 8).Value = "VO"
End Sub

Fama-MacBeth (1973) Two-Step procedure

MR Coef. St.Err. t-value p-value [95% Conf Interval] Sig
LMV .001 .002 0.34 .732 -.003 .004
PTBV2 .028 .019 1.48 .145 -.01 .066
PE 0 0 1.36 .18 0 0
EPS 0 0 1.83 .073 0 0 *
LVO .003 .001 3.73 0 .001 .005 ***
Constant -.03 .016 -1.85 .069 -.062 .002 *

Mean dependent var 0.005 SD dependent var 0.107

R-squared 0.101 Number of obs 5125

F-test 4.877 Prob > F 0.001

*** p<.01, ** p<.05, * p<.1

 LMV (Market Value): The coefficient in market value coefficient is 0.001. Moreover, the t-
statistics of 0.34 and the p-value of 0.732 indicate that the relationship between LMV and
the market return (PSX 100) is not statistically significant. The confidence intervals lie
between -0.003 to 0.004, indicating a week relationship between both variables.
 PTBV (Price-to-Book Value): The coefficient in market value PTBV is 0.028. Moreover, the t-
statistics of 1.48 and the p-value of 0.145 suggest that they are statistical significance The
confidence intervals lie between -0.01 to 0.066, indicating some unpredictability in the
estimate.

 PE (Price-to-Earnings): The coefficient for PE is 0, indicating no relationship between PE and

the market return. The t-value of 1.36 and the p-value of 0.18 suggest that this is statistically
insignificant.
  EPS (Earnings per Share): The coefficient for EPS is also 0, indicating no relationship between
EPS and market return. However, the t-value of 1.83 and the p-value of 0.073 suggest that
there is significancy but a weaker relationship. The confidence interval spans from 0 to 0.
 LVO (Volume): The coefficient for LVO is 0.003. The t-value of 3.73 and the p-value of 0
indicate a statistically significant positive relationship between LVO and market return. The
confidence interval ranges from 0.001 to 0.005.
 Constant: The constant is -0.03. The t-value of -1.85 and the p-value of 0.069 indicate that
the constant term is not statistically significant. The confidence interval ranges from -0.062
to 0.002. However, at 95% confidence interval, value must be less then -1.96 but in the case
of constant, the t-value is -1.85 which is statistically insignificant but among all variables, it is
the best result.
Overall, the model's R-squared value of 10.1% indicates that the independent variables (LMV,
PTBV, PE, EPS, LVO & Constant) explain 10.1% of the variation in the Market Returns. The F-test
of 4.877 with a p-value of 0.001 indicates that the model's is statistically significant. Among all
the variables, constant seems to be the best fit in the analysis and constant factor is the most
important one for returns.

Pairwise correlations
Variables (1) (2) (3) (4) (5) (6)
(1) MR 1.000
(2) LMV 0.032 1.000
(3) PTBV2 0.011 0.280 1.000
(4) PE 0.016 -0.018 0.006 1.000
(5) EPS 0.002 0.035 0.147 -0.003 1.000
(6) LVO 0.068 0.097 -0.395 -0.014 -0.083 1.000

 MR and LMV: The correlation between MR and LMV are 0.032, indicating a very weak
positive correlation. This suggests that there is a slight tendency for higher market values to
be associated with slightly higher market returns, but the relationship is weak positive.
Change in one variable will not cause change in another variable.
 MR and PTBV: The correlation coefficient between MR and PTBV2 are 0.011, indicating a very
weak correlation. This suggests that there is a minimal tendency for higher price-to-book
values to be associated with higher market returns. Change in one variable will not cause
change in another variable.
 MR and PE: The correlation coefficient between MR and PE is 0.016, indicating a very weak
positive correlation. This suggests that there is a slight tendency for higher price-to-earnings
ratios to be associated with slightly higher market returns. Change in one variable will not
cause change in another variable.
 MR and EPS: The correlation coefficient between MR and EPS is 0.002, indicating a very weak
positive correlation. However, among all the variables, EPS have the highest correlation with
market but still not statistically significant.
 MR and LVO: The correlation coefficient between MR and LVO is 0.068, indicating a very
weak positive correlation. This suggests that there is a slight tendency for higher trading
volume to be associated with slightly higher market returns, but the relationship is not
particularly statistical.
 Overall, the correlations between the variables are generally weak, indicating that the
relationships between these variables are not strongly corelated with Market Returns.However,
if were to choose any winner, the relations ship between MR and EPS have the lowest value of
0.002 which again is statistically insignificant.

Q4
Q4 A
Standard Daviation of
Beta 0.62
Average -2.73
Change -1.693

We can observe that on an average, the cross-sectional standard deviation is 0.62.

This is equal to -1.693 (0.62 x -2.73) when multiplied by the average slope from the
table in the preceding cell. This suggests that a one standard deviation variation in is
linked to a -1.693% annual decline in stock returns.

Q4 B
Outcome Variable 1 2 3 4 5 6 7 7-1
Average Betas -0.22 0.14 0.41 0.71 1.03 1.37 1.94 2.16
Average of Portfolio 0.96  
Average Betas*Avg of
-0.2112 0.1344 0.3936 0.6816 0.9888 1.3152 1.8624 2.0736
Portfolio

We effectively see the average return according to the stock's beta by multiplying the betas by the
average of portfolio. The return of each stock in relation to the overall market return is estimated
using this factor.
At portfolio1, which has a beta of -0.22, as an example. We get -0.2112 by multiplying 0.96. This
negative value shows that the stock is negatively associated with market movements and is
anticipated to have a lower return than the average market return.
In a similar manner, we multiply the average return by the corresponding betas of the other stocks
to get Average Betas*Avg of Portfolio.
Also, it seems that portfolios which are made since beta values with 1 being lowest betas and 7
being highest. There seems to be a direct relation between betas and expected returns; as beta
value is increasing the expected returns are increasing.
Also, the graph above shows that as our betas was increasing depending upon the portfolios, our
betas was increasing and average betas * average of portfolio/expected return kept on increasing
since our average portfolio is a positive value.

Q4C

According to the facts given, it seems that Regression 1's results are more statistically significant to
work with than Regression 4's. Here is the justification:

Greater absolute t-statistic in Regression 1, Beta 1's absolute t-statistic is 1.47, but in Regression 4, it
is 0.57. A stronger statistical correlation between the independent and dependent variables is
indicated by a larger absolute t-statistic. As a result, regression 1's is more statistically significant.

Lower p-value: In Regression 1, Beta 1's p-value is 0.16, which is considerably less than Regression
4's p-value of 0.57. A lower p-value denotes statistical significance that is larger. As a result,
Regression 1's is more statistically significant.

Regression 1 is statistically more significant to deal with than Regression 4 due to its greater
absolute t-statistic and lower p-value.