Unifying ESG Metrics and Market Performance for Biologic Industries

ESG
economics
healthcare
pharmaceuticals
Author

Bryan Armpriest

Published

March 2, 2025

Introduction and Background

Stock market data can be a critical indicator for how people feel about the company’s growth and future. How much a stock changes from year to year– regardless of day-to-day ups and downs–can be indicative of their current success or future promise.

Like most companies, those involving biological materials, pharmaceutical preparations, and In Vitro and In Vivo testing can be influenced by the risks that they are exposed to. In particular, risks However, these industries are likely separated for good reason, and will likely have different risk trends that impact how they do business. I want to expose these trends, and see if they play an impact on their annual market data. Combining market data and ESG ratings to see if there are general trends in industries can be a powerful tool to measure the impact of external aspects, and if the ESG risk score really matters when considering investments.

Research Question: Are annual stock prices of biology-related companies associated with their exposure to long-term risks from social, environmental, and governance sources? Particularly, do in-vitro/in-vivo diagnostic-producing companies have higher risk scores and therefore lower stock prices? I also want to take a look at controversy scores for each of these companies.

from google.colab import drive
drive.mount('/content/drive')
Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount("/content/drive", force_remount=True).

#Data Collection

Initially, I was provided with esg_proj_2024_data, and esg_proj_2025.

Using selenium in the Spyder IDE, I was able to obtain and clean two DataFrames from Yahoo! Finance. The first, was a DataFrame involving stocks, their prices, and volatility. This data was collected using a table-scraping method on Yahoo! Finance’s website and stored a DataFrame: market_df.

The second DataFrame was also from Yahoo! Finance, and included ESG Risk Ratings for each company in the esg_proj_2024 dataframe.

To begin, I need to merge some of the dataframes together, and clean a bit more to make the data workable. For this project, I will be honing in on 2025 market data, and the most recent ESG Risk data.

import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
url_2025 = "https://bcdanl.github.io/data/esg_proj_2025.csv"
esg_proj_2025 = pd.read_csv(url_2025)
esg_proj_2025.head()
Year Symbol Name Sector Industry Country Market_Cap IPO_Year
0 2025 A Agilent Technologies Inc. Common Stock Industrials Biotechnology: Laboratory Analytical Instruments United States 3.391867e+10 1999.0
1 2025 A Agilent Technologies Inc. Common Stock Industrials Biotechnology: Laboratory Analytical Instruments United States 3.391867e+10 1999.0
2 2025 AA Alcoa Corporation Common Stock Industrials Aluminum United States 8.279121e+09 2016.0
3 2025 AA Alcoa Corporation Common Stock Industrials Aluminum United States 8.279121e+09 2016.0
4 2025 AAL American Airlines Group Inc. Common Stock Consumer Discretionary Air Freight/Delivery Services United States 7.325392e+09 NaN 
url_2024 = "https://bcdanl.github.io/data/esg_proj_2024_data.csv"
esg_proj_2024_data = pd.read_csv(url_2024)

esg_proj_2024_data['Year'] = 2024
esg_proj_2024_data.head()
Year Symbol Name Sector Industry Country Market_Cap IPO_Year Total_ESG Environmental Social Governance Controversy
0 2024 A Agilent Technologies Inc. Common Stock Industrials Biotechnology: Laboratory Analytical Instruments United States 40365434818 1999.0 13.6 1.1 6.4 6.1 2.0
1 2024 AA Alcoa Corporation Common Stock Industrials Aluminum United States 6622135551 2016.0 24.0 13.8 5.9 4.3 3.0
2 2024 AAL American Airlines Group Inc. Common Stock Consumer Discretionary Air Freight/Delivery Services United States 9088024606 NaN 26.4 9.9 11.6 4.8 2.0
3 2024 AAP Advance Auto Parts Inc. Consumer Discretionary Auto & Home Supply Stores United States 4474665296 NaN 11.5 0.1 8.3 3.1 2.0
4 2024 AAPL Apple Inc. Common Stock Technology Computer Manufacturing United States 2614310000000 1980.0 17.2 0.5 7.4 9.4 3.0

Loading the dataframes that I’ve collected using selenium from Yahoo! Finance

market_df = pd.read_csv('/content/drive/My Drive/market_clean.csv')
market_df.head()
DtypeWarning: Columns (7) have mixed types. Specify dtype option on import or set low_memory=False.
  market_df = pd.read_csv('/content/drive/My Drive/market_clean.csv')
Unnamed: 0 Date Open High Low Close Adj Close Volume Symbol Year
0 0 2025-03-28 119.21 119.66 116.36 116.69 116.44 1772900 A 2025
1 1 2025-03-27 120.00 120.33 118.73 118.97 118.72 2075700 A 2025
2 2 2025-03-26 120.72 121.14 119.05 119.90 119.65 2652900 A 2025
3 3 2025-03-25 122.60 123.04 119.76 120.51 120.25 1605900 A 2025
4 4 2025-03-24 121.88 124.43 120.89 121.97 121.71 1557400 A 2025 
market_df_2025 = market_df.query("Year == 2025")
market_df_2025
Unnamed: 0 Date Open High Low Close Adj Close Volume Symbol Year
0 0 2025-03-28 119.21 119.66 116.36 116.69 116.44 1772900 A 2025
1 1 2025-03-27 120.00 120.33 118.73 118.97 118.72 2075700 A 2025
2 2 2025-03-26 120.72 121.14 119.05 119.90 119.65 2652900 A 2025
3 3 2025-03-25 122.60 123.04 119.76 120.51 120.25 1605900 A 2025
4 4 2025-03-24 121.88 124.43 120.89 121.97 121.71 1557400 A 2025
... ... ... ... ... ... ... ... ... ... ...
206502 209101 2025-01-08 30.51 30.70 30.21 30.68 30.68 1048000 MFC 2025
206503 209102 2025-01-07 30.90 31.03 30.47 30.66 30.66 971900 MFC 2025
206504 209103 2025-01-06 31.09 31.25 30.76 30.81 30.81 999300 MFC 2025
206505 209104 2025-01-03 30.74 30.82 30.53 30.67 30.67 824500 MFC 2025
206506 209105 2025-01-02 30.75 30.82 30.31 30.52 30.52 872500 MFC 2025

21771 rows × 10 columns

annual_summary_2025 = market_df_2025.groupby(['Symbol']).agg(
    first_close = ('Close', 'first'),
    last_close = ('Close', 'last')
).reset_index()

annual_summary_2025[['first_close', 'last_close']] = annual_summary_2025[['first_close', 'last_close']].astype('float')

annual_summary_2025['Annual Total Return'] = annual_summary_2025['last_close']/(annual_summary_2025['first_close'] - 1)
annual_summary_2025.head() #This is going to be a very nice, clean DataFrame to work with
Symbol first_close last_close Annual Total Return
0 A 116.69 133.43 1.153341
1 AA 30.66 37.99 1.280850
2 AAL 10.70 17.00 1.752577
3 AAP 38.95 48.14 1.268511
4 AAPL 217.90 243.85 1.124251 
esg_df = pd.read_csv('/content/drive/My Drive/project_data_2.csv')
esg_df.head()
Total_ESG Environmental Social Governance Controversy Symbol
0 10.1 1.1 5.0 3.9 1.0 A
1 25.1 14.7 8.0 2.4 3.0 AA
2 23.8 9.6 11.3 2.9 2.0 AAL
3 10.7 0.2 8.2 2.3 2.0 AAP
4 18.8 2.0 8.4 8.4 3.0 AAPL 
# Allow me to combine the ESG DataFrame with the esg_proj_2025 DataFrame
esg_merged = pd.merge(esg_df, annual_summary_2025, on = 'Symbol', how = 'inner')
esg_merged #I'm unsure where all of the NaN values are coming from here.
Total_ESG Environmental Social Governance Controversy Symbol first_close last_close Annual Total Return
0 10.1 1.1 5.0 3.9 1.0 A 116.69 133.43 1.153341
1 25.1 14.7 8.0 2.4 3.0 AA 30.66 37.99 1.280850
2 23.8 9.6 11.3 2.9 2.0 AAL 10.70 17.00 1.752577
3 10.7 0.2 8.2 2.3 2.0 AAP 38.95 48.14 1.268511
4 18.8 2.0 8.4 8.4 3.0 AAPL 217.90 243.85 1.124251
... ... ... ... ... ... ... ... ... ...
344 22.0 9.3 9.0 3.7 3.0 MDLZ 67.72 59.50 0.891787
345 18.7 2.0 10.3 6.4 2.0 MDT 87.63 80.22 0.926007
346 15.1 0.5 7.2 7.4 2.0 MET 79.23 81.78 1.045379
347 32.7 2.4 21.6 8.7 4.0 META 576.74 599.24 1.040817
348 14.9 1.1 7.5 6.2 2.0 MFC 30.76 30.52 1.025538

349 rows × 9 columns

esg_proj_2025['Industry'].unique() #I am going to use this line to get a feel for what industries to hone in on.
esg_proj_2025_filtered = esg_proj_2025[['Symbol', 'Industry']].query(
    "Industry == 'Biotechnology: Biological Products (No Diagnostic Substances)' | Industry == 'Biotechnology: In Vitro & In Vivo Diagnostic Substances' | Industry == 'Biotechnology: Pharmaceutical Preparations'")

esg_proj_2025_filtered.head()

esg_proj_2025_filtered['Industry'].value_counts()

## There was an oddly low number of companies under some of the industries of interest.
# There are only 7 companies listed that are involved in diagnostic substances, but there is an abundance of pharmaceutical companies.
count
Industry
Biotechnology: Pharmaceutical Preparations 215
Biotechnology: Biological Products (No Diagnostic Substances) 77
Biotechnology: In Vitro & In Vivo Diagnostic Substances 7 

#Finally, I want to merge the DataFrame that I just created to the previously merged DataFrame so filter out the industries that we want.
df_all = pd.merge(esg_proj_2025_filtered, esg_merged, on = 'Symbol', how = 'inner')
df_all.head() #Looks good!
df_all['Industry'].value_counts() #We are now down to 16, 4, and 1 companies for the
count
Industry
Biotechnology: Pharmaceutical Preparations 9
Biotechnology: Biological Products (No Diagnostic Substances) 3
Biotechnology: In Vitro & In Vivo Diagnostic Substances 1

Descriptive Statistics

esg_merged[['Total_ESG', 'Environmental', 'Social', 'Governance', 'Controversy', 'Annual Total Return']].describe()
#Important metrics can be gathered from the more general statistics.
#The average ESG average score is about 20.8, the average controversy score is 1.89, and the average annual total return is about 1.05 (slightly positive)
Total_ESG Environmental Social Governance Controversy Annual Total Return
count 349.000000 349.000000 349.000000 349.000000 324.000000 349.000000
mean 20.603152 6.021777 9.020344 5.559885 1.904321 1.053716
std 6.894848 5.503169 3.518639 2.511559 0.734033 0.150385
min 5.700000 0.000000 0.900000 1.400000 1.000000 0.668582
25% 15.300000 1.800000 6.800000 3.900000 1.000000 0.946971
50% 20.400000 4.000000 9.000000 4.900000 2.000000 1.041594
75% 24.900000 9.400000 11.300000 6.400000 2.000000 1.128365
max 45.200000 31.100000 22.200000 14.600000 4.000000 1.878179 
#Distribution of General ESG Data
(
 sns.FacetGrid(
       data = esg_merged)
 .map(sns.histplot, 'Total_ESG')
 )

(
 sns.FacetGrid(
       data = esg_merged)
 .map(sns.histplot, 'Annual Total Return')
 )

#Next, let's take a look at these three groups in the categories that we wanted to take a look into.
df_all[['Symbol', 'Total_ESG', 'Environmental', 'Social', 'Governance', 'Controversy', 'Annual Total Return']].describe()
Total_ESG Environmental Social Governance Controversy Annual Total Return
count 13.000000 13.000000 13.000000 13.000000 13.000000 13.000000
mean 22.761538 1.884615 14.400000 6.469231 2.230769 0.979758
std 3.860384 1.335319 2.744995 0.945502 1.012739 0.169004
min 16.300000 0.400000 7.700000 5.100000 1.000000 0.829317
25% 20.500000 0.900000 13.000000 6.000000 1.000000 0.878359
50% 22.500000 1.800000 14.600000 6.200000 2.000000 0.943924
75% 23.600000 2.400000 16.000000 7.000000 3.000000 0.985489
max 32.500000 5.500000 18.700000 8.200000 4.000000 1.448905 
#Distribution of all three industries in question
(
 sns.FacetGrid(
       data = df_all,
       row='Industry')
 .map(sns.histplot, 'Total_ESG')
 )

(
 sns.FacetGrid(
       data = df_all,
       row='Industry')
 .map(sns.histplot, 'Controversy')
 )

(
 sns.FacetGrid(
       data = df_all,
       row='Industry')
 .map(sns.histplot, 'Annual Total Return')
 )

Questions to Answer (post-EDA)

From above, I have copied the three questions that I had wanted to explore. However, I don’t believe that I will be able to produce meaningful results for the second question, since the sample size is miniscule.

  1. Are annual stock prices of biology-related companies associated with their exposure to long-term risks from social, environmental, and governance sources?

  2. Particularly, do in-vitro/in-vivo diagnostic-producing companies have higher risk scores and therefore lower stock prices?

  3. I also want to take a look at controversy scores for each of these companies, and relate it to their annual return.

#1. Is the annual stock return of biology-related companies associated with their exposure to long-term risks from social, environmental, and governance sources?
#df_all.columns
#There were incredibly limited numbers for the companies highlighted in the ESG 2024 data.
sns.lmplot(data = df_all,
           x = 'Total_ESG',
           y = 'Annual Total Return')

There seems to be a generally positive relationship between the ESG Risk indicator and the Annual Total Return of bio-related industries. However, it doesn’t seem like it’s a very clear relationship, and there is limitation in the data size. Based on the total return, it seems like there is a decrease in the return of many of these companies. Based on the consensus of shareholders, it seems like there is going to be a decline in other metrics, such as profitablity of these companies. In comparison to the average ESG value (about 20), most companies are above that threshold, with one particular company that could be skewing the results of the correlation.

Answer: ESG Risk Score seems to be positively correlated with bio-industry company annual stock return for 2025. This seems odd, seeing how the risks from different sources could be seen as risky investments. But then again, as many say, without risk, there is no reward.

Next, I want to look at the third question I had, which was relating the controversy scores of each bio-related company to their annual total return.

sns.lmplot(data = df_all,
           x = 'Controversy',
           y = 'Annual Total Return')

sns.lmplot(data = df_all,
           x = 'Controversy',
           y = 'Annual Total Return')

There are outliers in the DataFrame of controversy levels of bio-industry companies, and there seems to be no relationship between controversy and annual return.

Answer: The relationship between controversy and annual return is ambiguous. There are controversial companies that seem to be successful, whereas there are non-controversial companies that aren’t so successful.

Closing Remarks

Limited by data size, and collection time, I was unable to find conclusive results beside the fact that there was a positive relationship between Annual Total Return and ESG risk score in 2025.

This is an important finding, and could be applied to the understanding of certain stocks. In the biological fields and companies that relate to them, there seems to be a reward for those that are most risky. Creating products that are subject to more governmental, social, and environmental risks can either be beneficial for the business, or lead to loss of favor from stockholders.

Citations

Yahoo Finance. (n.d.). General stock market data. Retrieved May 15, 2025, from https://finance.yahoo.com

Selenium. (n.d.). SeleniumHQ Browser Automation. https://www.selenium.dev