META’s Secret in ESG issues (Spring 2022)
[Notice] Journey to the academic researcher This is the story of how I became the insightful researcher.
META’s Secret in ESG issues
Among 10 companies which experienced two-notch downgrade, only two companies became laggard ones. I wanted to point out Meta Platforms because the stock price of Meta plummeted by 24.15% in a day.
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
df=pd.read_csv('msci_timeseries_20210901.csv')
fb = df.query('ISSUER_ISIN == "US30303M1027"')
fb.reset_index(drop=True, inplace=True)
fb = fb.loc[0, :]
fb
ISSUER_NAME FACEBOOK, INC.
ISSUERID IID000000002638948
ISSUER_TICKER FB
ISSUER_CUSIP 30303M102
ISSUER_SEDOL B7TL820
...
CORP_GOVERNANCE_GOV_PILLAR_SD -3.3
ACCOUNTING_GOV_PILLAR_SD -0.29
BOARD_GOV_PILLAR_SD -0.86
OWNERSHIP_GOV_PILLAR_SD -1.41
PAY_GOV_PILLAR_SD -0.78
Name: 0, Length: 206, dtype: object
finder = fb.filter(regex = 'WEIGHT')
weight = finder[(finder > 0)]
weight
WEIGHTED_AVERAGE_SCORE 3.7
ENVIRONMENTAL_PILLAR_WEIGHT 5
SOCIAL_PILLAR_WEIGHT 53
GOVERNANCE_PILLAR_WEIGHT 42
CLIMATE_CHANGE_THEME_WEIGHT 5.0
HUMAN_CAPITAL_THEME_WEIGHT 24
PRODUCT_SAFETY_THEME_WEIGHT 29.0
CARBON_EMISSIONS_WEIGHT 5.0
HUMAN_CAPITAL_DEV_WEIGHT 24.0
PRIVACY_DATA_SEC_WEIGHT 29.0
Name: 0, dtype: object
inter_ind = df[(df['IVA_INDUSTRY'] == 'Interactive Media & Services')]
inter_ind[['WEIGHTED_AVERAGE_SCORE','INDUSTRY_ADJUSTED_SCORE']].describe().round(3)
| WEIGHTED_AVERAGE_SCORE | INDUSTRY_ADJUSTED_SCORE | |
|---|---|---|
| count | 23.000 | 23.000 |
| mean | 4.104 | 4.391 |
| std | 0.952 | 2.296 |
| min | 2.100 | 0.800 |
| 25% | 3.450 | 2.850 |
| 50% | 4.100 | 4.100 |
| 75% | 4.750 | 6.050 |
| max | 6.500 | 10.000 |
inter_ind['IVA_COMPANY_RATING'].value_counts()[['CCC','B','BB','BBB','A','AA','AAA']].plot(kind='bar')
count = inter_ind['IVA_COMPANY_RATING'].value_counts()
count
BB 6
B 5
BBB 4
A 4
AA 2
AAA 1
CCC 1
Name: IVA_COMPANY_RATING, dtype: int64
I used Bottom-up analysis based on MSCI ESG rating methodology. I thought four numbers are extreme. The first one is Human Capital Development. Meta’s exposure score is high, but its management seems not enough. However, I figured out the exposure scores of companies in Interactive Media & Services are generally high as Meta’s and Meta’s management score is higher than industry’s average. Like HCD, Ownership and Control subjection points is a little bit lower than industry’s average, but it is not extremely bad.
hcd_ind = inter_ind['HUMAN_CAPITAL_DEV_MGMT_SCORE']
plt.hist(hcd_ind, bins=7)
(array([1., 0., 3., 1., 3., 8., 7.]),
array([0.7 , 1.42857143, 2.15714286, 2.88571429, 3.61428571,
4.34285714, 5.07142857, 5.8 ]),
<BarContainer object of 7 artists>)
hcd_ind.describe()
count 23.000000
mean 4.334783
std 1.245942
min 0.700000
25% 4.100000
50% 4.600000
75% 5.100000
max 5.800000
Name: HUMAN_CAPITAL_DEV_MGMT_SCORE, dtype: float64
hcd_ind = inter_ind['HUMAN_CAPITAL_DEV_EXP_SCORE']
plt.hist(hcd_ind, bins=7)
(array([ 1., 0., 1., 1., 14., 1., 5.]),
array([6.2 , 6.64285714, 7.08571429, 7.52857143, 7.97142857,
8.41428571, 8.85714286, 9.3 ]),
<BarContainer object of 7 artists>)
hcd_ind.describe()
count 23.000000
mean 8.343478
std 0.699887
min 6.200000
25% 8.200000
50% 8.300000
75% 8.450000
max 9.300000
Name: HUMAN_CAPITAL_DEV_EXP_SCORE, dtype: float64
hcd = df['HUMAN_CAPITAL_DEV_MGMT_SCORE']
plt.hist(hcd, bins=10)
(array([ 74., 96., 158., 291., 427., 344., 321., 303., 155., 48.]),
array([0. , 0.92, 1.84, 2.76, 3.68, 4.6 , 5.52, 6.44, 7.36, 8.28, 9.2 ]),
<BarContainer object of 10 artists>)
hcd.describe()
count 2217.000000
mean 4.800496
std 1.913621
min 0.000000
25% 3.500000
50% 4.800000
75% 6.300000
max 9.200000
Name: HUMAN_CAPITAL_DEV_MGMT_SCORE, dtype: float64
hcd = df['HUMAN_CAPITAL_DEV_EXP_SCORE']
plt.hist(hcd, bins=10)
(array([115., 84., 164., 145., 292., 277., 452., 325., 222., 141.]),
array([ 2.3 , 3.07, 3.84, 4.61, 5.38, 6.15, 6.92, 7.69, 8.46,
9.23, 10. ]),
<BarContainer object of 10 artists>)
hcd.describe()
count 2217.000000
mean 6.673207
std 1.833972
min 2.300000
25% 5.500000
50% 7.000000
75% 8.100000
max 10.000000
Name: HUMAN_CAPITAL_DEV_EXP_SCORE, dtype: float64
Meta’s Privacy Data Security exposure score is higher than industry’s average. However, Meta’s management score is lower than industry’s average. Therefore, we can infer that the gap between the exposure and management triggered Meta to get a bad score in this issue.
pds_ind = inter_ind['PRIVACY_DATA_SEC_EXP_SCORE']
plt.hist(pds_ind, bins=7)
(array([1., 0., 2., 2., 9., 5., 4.]),
array([5.3 , 5.92857143, 6.55714286, 7.18571429, 7.81428571,
8.44285714, 9.07142857, 9.7 ]),
<BarContainer object of 7 artists>)
pds_ind.describe()
count 23.000000
mean 8.234783
std 0.982394
min 5.300000
25% 7.900000
50% 8.200000
75% 8.850000
max 9.700000
Name: PRIVACY_DATA_SEC_EXP_SCORE, dtype: float64
pds_ind = inter_ind['PRIVACY_DATA_SEC_MGMT_SCORE']
plt.hist(pds_ind, bins=7)
(array([2., 6., 9., 2., 2., 0., 2.]),
array([3.3 , 4.11428571, 4.92857143, 5.74285714, 6.55714286,
7.37142857, 8.18571429, 9. ]),
<BarContainer object of 7 artists>)
pds_ind.describe()
count 23.000000
mean 5.421739
std 1.416448
min 3.300000
25% 4.500000
50% 5.200000
75% 5.850000
max 9.000000
Name: PRIVACY_DATA_SEC_MGMT_SCORE, dtype: float64
pds = df['PRIVACY_DATA_SEC_MGMT_SCORE']
plt.hist(pds, bins=10)
(array([166., 74., 101., 192., 339., 661., 333., 191., 86., 22.]),
array([ 0., 1., 2., 3., 4., 5., 6., 7., 8., 9., 10.]),
<BarContainer object of 10 artists>)
pds.describe()
count 2165.000000
mean 4.885820
std 2.079936
min 0.000000
25% 4.200000
50% 5.000000
75% 6.300000
max 10.000000
Name: PRIVACY_DATA_SEC_MGMT_SCORE, dtype: float64
pds = df['PRIVACY_DATA_SEC_EXP_SCORE']
plt.hist(pds, bins=10)
(array([ 12., 127., 746., 272., 144., 139., 222., 222., 205., 75.]),
array([0.5 , 1.42, 2.34, 3.26, 4.18, 5.1 , 6.02, 6.94, 7.86, 8.78, 9.7 ]),
<BarContainer object of 10 artists>)
pds.describe()
count 2164.000000
mean 4.696026
std 2.220830
min 0.500000
25% 2.700000
50% 3.900000
75% 6.800000
max 9.700000
Name: PRIVACY_DATA_SEC_EXP_SCORE, dtype: float64
ogp_ind = inter_ind['OWNERSHIP_GOV_PILLAR_SD']
plt.hist(ogp_ind, bins=7)
(array([4., 3., 3., 3., 2., 2., 6.]),
array([-2.3 , -1.98285714, -1.66571429, -1.34857143, -1.03142857,
-0.71428571, -0.39714286, -0.08 ]),
<BarContainer object of 7 artists>)
ogp_ind.describe()
count 23.000000
mean -1.113043
std 0.736697
min -2.300000
25% -1.780000
50% -1.250000
75% -0.410000
max -0.080000
Name: OWNERSHIP_GOV_PILLAR_SD, dtype: float64
As you can see this graph, in the Business Ethics issue, Meta is the one of the worst companies in the industry. As a result, I figured out Two Issues, Privacy Data Security and Business Ethics, made Meta be an ESG laggard.
ogp = df['OWNERSHIP_GOV_PILLAR_SD']
plt.hist(ogp, bins=10)
(array([ 1., 1., 2., 5., 34., 75., 209., 524., 590., 774.]),
array([-4.06 , -3.654, -3.248, -2.842, -2.436, -2.03 , -1.624, -1.218,
-0.812, -0.406, 0. ]),
<BarContainer object of 10 artists>)
ogp.describe()
count 2215.000000
mean -0.709819
std 0.501394
min -4.060000
25% -1.020000
50% -0.620000
75% -0.270000
max 0.000000
Name: OWNERSHIP_GOV_PILLAR_SD, dtype: float64
I think Privacy data security and Business ethics are not mutually exclusive. Basically Privacy data is users basic right, poorly managing pds can mean Meta’s business ethics are wrong. Because Meta has 2.91 billion Monthly Active users, Meta should establish the most robust security system. However, since the Scandal in 2018, Hackers have penetrated Meta’s database and steal users’ data even in 2021. In the Business Ethics aspect. Meta confronts a variety of issues. One of the crucial issues is Meta collects data without users’ consent. Pointing out this issue, last year Apple started to provide a new service preventing sharing users’ data on social media. The service encouraged users to realize how badly Meta collects users’ data. Especially yesterday, Mark Zuckerberg threatened to pull over Facebook and Instagram in Europe against data privacy regulation rather than comply with it. Meta behaves like a tech dictator.
be_ind = inter_ind['BUS_ETHICS_GOV_PILLAR_SD']
plt.hist(be_ind, bins=7)
(array([ 3., 2., 3., 2., 11., 1., 1.]),
array([-3.3 , -2.94285714, -2.58571429, -2.22857143, -1.87142857,
-1.51428571, -1.15714286, -0.8 ]),
<BarContainer object of 7 artists>)
be_ind.describe()
count 23.00000
mean -2.06087
std 0.62794
min -3.30000
25% -2.40000
50% -1.80000
75% -1.70000
max -0.80000
Name: BUS_ETHICS_GOV_PILLAR_SD, dtype: float64
be = df['BUS_ETHICS_GOV_PILLAR_SD']
plt.hist(be, bins=10)
(array([ 1., 9., 32., 127., 194., 410., 713., 530., 160., 44.]),
array([-4.5 , -4.07, -3.64, -3.21, -2.78, -2.35, -1.92, -1.49, -1.06,
-0.63, -0.2 ]),
<BarContainer object of 10 artists>)
be.describe()
count 2220.000000
mean -1.762027
std 0.623763
min -4.500000
25% -2.100000
50% -1.700000
75% -1.300000
max -0.200000
Name: BUS_ETHICS_GOV_PILLAR_SD, dtype: float64
Eventually, Meta’s destructive behaviors made users leave Facebook, and Meta’s earnings decreased. After Meta’s Q4 earnings announcement, the stock price plummeted. Handling poorly at two issues caused not only Meta’s CF to decrease but also future risk to increase. Investors concerned about the ESG rating would avoid investing in Meta if they checked that the rating plummeted last year.