Aboli Khairnar, Data Scientist, Citi Ventures Inc. The majority of traditional corporate valuation methods are solely based on tangible indicators ' sales growth, gross profit, cash flow, and operational performance ' which are the result and don't reflect the underlying process which makes the organization successful in the longer run. One of the most important intangible assets that don't appear directly on the balance sheet is human capital. In this research, we focus on the role that workforce skills, education, and knowledge play towards organizational success using state-of-the-art Machine Learning techniques. We find that investments in human capital not only play a crucial role in organizational growth but also have a causal relationship.

1. Human Capital Growth Analytics
DataconLA | August 2022
Citi Ventures| Venture Innovation

2. Intangible assets of an enterprise that are required to
achieve business goals, including employee's
knowledge; data and information about processes,
products, customers and competitors; and intellectual
property such as patents or regulatory licenses.
(Source: Gartner)
Intellectual/ Knowledge Capital
Jul 14, 2022

3. Intellectual/ Knowledge Capital
Intellectual Capital
Human capital Relational capital Structural capital
• The contributions made to an organization
by its employees utilizing their talents,
skills, and expertise.
• Possessed only by individuals but may be
harnessed by an organization.
• Provides a comparative advantage
• Quality organizations are ones that focus
on retaining creative and innovative
workers, as well as work toward creating a
setting where such intelligence can be
taught and learned.
• The relationships between
coworkers as well as those
between workers and suppliers,
customers, partners, and
collaborators.
• Relationship capital also includes
franchises, licenses, and
trademarks as they have value
only in the context of the
relationship they have with
customers.
• The non-physical capital possessed
by an organization—such as
processes, method, and techniques—
that allow it to operate and enable it to
leverage its capabilities.
• Structural capital may include
intellectual property such as
databases, code, patents, proprietary
processes, trademarks, software, and
more.

4. Methodology

5. Assumptions
5
Source: Data entry best practices
Data Accuracy
Source: Data entry best practices
No Spillover Effects for R&D Data Missing at Random
Constant Workplace Productivity
Source: Aimstyle
Source: Towards Data Science
Source: Getty Images

6. Data Science Workflow
6
Data Preprocessing
Modeling
Recommendations
Financial data
Employee data
Data Cleaning
Classification using
Random Forest
Human Capital(HC)
Index1
Customized HC
investment2
1. Index is based on the probability which is calculated using importance of each feature as well as SHAP values which depicts how that feature affects the output in presence of other factors. In
other words, it is a number between 0-100, calculated based on the model output which is also the probability of firm being a high growth firm (a firm having growth rate > 2.5%). If our model
predicts the probability output as 0.5, then the Human Capital Index for that firm will be (100*0.5) = 50
2. Customized HC investments are based on SHapely Additive exPlanations (SHAP) values for those human capital factors. The goal of customized investments is to highlight critical human
capital factors along with their optimum allocations aimed to increase the total revenue growth rate for a firm

7. Data Sources1
Employee data Financial data
Salary, Attrition, Gender, Race, Job Titles, Skills etc. Revenue, Profitability, Return on Equity, Stock Price etc.
Missing Data
– Majority of firms in the financial sector don’t report R&D
– Around 16% of R&D data is filled-in using various computing techniques2
Data: ~2150 US firms
7
1. Employee and financial data are provided by Revelio Labs and S&P financials database Capital IQ respectively

8. Classification Using H20 AutoML
R&D Expense
as of Total Revenue
Location
Firm Size
Sector
Employees across top
Skill Clusters
Employees in
Technical roles
Is it a High Growth firm?
• SAP/ Accounting/ Financial Reporting
• SQL/ Linux/ Development
• Marketing/ Social Media/ Marketing Strategy
• Data Analysis/ Databases/ Data Warehousing
• Java/ C++/ C
• Analysis/ Financial Analysis/ Finance
• Human Resources/ Recruiting/ Performance
Management/
• Program Management/ Security/ Engineering
management
• Information technology/ people management/
project management
Project Manager, Database Administrator, Test
Engineer, Data Analyst, Graphic Designer,
Software Developer, IT Project Manager, Web
Developer, Application Engineer, Infrastructure
Engineer, Information Security, Quality Engineer,
Scientist, Data Engineer, UX Designer, Systems
Engineer, Software Engineer, Data Scientist,
Business Analyst, Automation Engineer,
Technology Lead, IT Analyst, DevOps Engineer,
Product Manager, Technical Support Engineer,
Technology Analyst, Technical Architect

9. Key Results

10. 10
1. Data Skills: Skills specific to Data Science/ Data Engineering roles; Software Development Skills: Skills specific to Software Engineering/ Developer roles
2. Partial Dependence Plot (PDP) shows marginal effect of a variable in mean response by assuming independence between feature for which PDP is calculated and the rest
3. Follows law of diminishing marginal utility – marginal utility increases at decreasing rate
4. X-axis description:
• For Data & Analytics, Software Development, Engineering Management skill clusters, X axis refers to % of employees in that skill cluster
• For R&D, it refers to %R&D expense as of Total Revenue
Key factors and their link to firms’ financial performance
Partial Dependence Plots (PDP2)
– Shows how each human
capital factor affects average
prediction about firm’s
performance
– Employees with Data1 skills
shows initial increase3 the
average output probability of
firm being a high growth firm
Probability
of
firm
being
high
growth
Employees within Technical and Engineering Management skill clusters along
with R&D increase the average probability of a firm being a high growth
Relationship between Human Capital Factors and firms' Financial Performance
% Workforce/Spend4

11. Company % R&D
% Tech
roles
% Employees by Skill Clusters
Sector State
HC
Index
Software
Development
Data
Analytics
Engineering
Management
A 42 50 87 28 15
Information
Technology
California 100
B 20 48 67 18 10
Information
Technology
California 97
C 20 48 65 15 15
Information
Technology
Massachusetts 97
D 21 49 54 15 15
Information
Technology
New York 91
E 19 43 9 14 5 Health care California 91
F 11 32 9 12 5 Health care Massachusetts 66
G 11 17 9 4 3
Communication
Services
California 54
H 1 12 2 3 2 Consumer Staples Michigan 22
I 3 19 4 4 4
Consumer
Discretionary
Michigan 20
J 1 16 5 6 5 Financials Massachusetts 18
K 2 15 5 3 3
Consumer
Discretionary
Texas 13
L 2 20 8 7 5 Financials New York 9
M 1 8 0 1 3 Industrials Florida 6
Human Capital Growth Index – Sample Output
The following table ranks a subset of companies in the US based on their HC index.
• The top-rated firm from the predicted index belongs to the IT sector ad located in California.
• The lowest rated firm belongs to the Industrials sector and is in Florida.
• Overall, the index values are specifically high for IT and Health care firms with High R&D and Tech workforce.
High
Growth
Low
Growth

12. Moderate correlation between Human Capital Index and Revenue Growth Rate
Human
Capital
Index
Total Revenue Growth Rate
Correlation 54%*
*Shown in percentage (0.54*100).
Ranges for Correlation Coefficient:
0.9 - 1 (Very high); 0.7 - 0.9 (High); 0.5 - 0.7 (Moderate); 0.3 - 0.5 (Low); 0 - 0.3 (Very Low)
Source: https://towardsdatascience.com/eveything-you-need-to-know-about-interpreting-correlations-2c485841c0b8

13. % IT/ People/Project Management
% Technical workforce
% HR/ Performance Management
% Program/ Engineering management
% R&D
% Java/ C/ C++
% SQL/ Linux/ Software Development
% Data Analysis/ Data Engineering
Contribution of Human Capital Factors towards Revenue
Growth Rate
BottomUp Solutions Vs. Apexify Labs
Compared to Apexify Labs, BottomUp Solutions continues to underutilize its
technical workforce and R&D spend
13
Case study
BottomUp Solutions Vs. Apexify Labs
Total Revenue Growth Rate (CY 2020)
– BottomUp Solutions: -5.5 %
– Apexify Labs : 12%
– For BottomUp Solutions, contributions1 of
all selected human capital factors is
consistently lower compared to that of
Apexify Labs
– The difference is significant2 for R&D
Expense and % employees across
Technical roles, Software Engineering,
Data, and Engineering Management skills
clusters.
– As per our hypothesis, there is scope to
improve BottomUp Solutions’s revenue
growth rate by increasing the
contribution of some of these human
capital factors
Note: Apexify Labs and BottomUp Solutions used in this presentation represent two fictional companies in the IT industry
1. Contributions can also be interpreted as weights which are used in calculating output probability of a firm being a high growth firm
2. Assumption: Significant where contribution for BottomUp < 0.5*(Apexify)
Apexify Solutions does an excellent job in creating value from their technical workforce
Impact on financial performance
% Data Analytics
% Software Development
% Engineering Management
Apexify
Labs
BottomUp
Solutions

14. Opportunities

15. Interactive app takes new allocations for
firm BottomUp Solutions’s key human
capital metrics as an input and predicts
if this firm would have been a high
growth firm with these new allocations.
15
Note: Above app takes following inputs for 2017-2019: % Technical workforce, % employees across SQL/ Linux/ Software Development, % employees across % Java/ C/ C++ and %R&D Expense
as of Total Revenue for 2017 as an input and predicts if firm would have been a high growth firm or not in 2020.
App Screenshot
Tech workforce
Software Development
Java
R&D
36
20
Human Capital Factors Existing Allocations (%)
8
8
New Allocations (%)
35
20
4
10
Run
Total Revenue Growth Rate with existing allocations is –5.5%
With new allocations BottomUp Solutions could have been a LOW growth firm. i.e. It’s predicted
Total Revenue Growth Rate < 2.5%
Understanding Human Capital Analytics can help the organization to
proactively plan its Human Capital decisions, which can lead to positive
financial performance

16. 36% 45%
Female Board Members
Dashboard view of metrics and scenario planning decisions
With a few changes BottomUp Solutions could have increased growth rate by 8%
16
Recommendation
One of the ways BottomUp Solutions might
have increased its revenue growth rate from
–5.5% to at least 2.5% is by making following
changes to its knowledge-based capital
– Increasing % employees with SQL/
Linux/ Software Development and Java/
C/ C++ skills by 4%
– Increasing %R&D spend by 5%
*Currently focusing on only certain aspects of
boosting productivity – hiring additional
employees with technical skills and
increasing R&D spend
Technical Skills
Software Development
Java
Research & Development
36% 40%
8% 12%
8% 13%
Note: The above combination shows only selected actionable human capital factors whose allocations are obtained by trial and error. Please note it represents one of the all-possible combinations
that could have classified firm as a high growth firm.

17. Next Steps

18. Enhance the Human Capital Index
– Explore better performing models that use alternative financial metrics such as EBITDA, total revenue, etc.
– Create new features to better explain our data
– Improve the interpretability of continuous Human Capital Index by creating distinct subcategories
– Incorporate new data, historical data, global data, data on startups, D&I
– Look into Causality
Next Steps
Overall, we can improve our human capital index and make it more robust by increasing number of firms and
adding new features to our model
18

19. Aboli Khairnar
aboli.khairnar@citi.com
Questions?

20. Thank you!