Social Impacts of Robotics on the Labor and Employment Market

A master's capstone project by Kelvin Espinal.
Submitted to the Graduate Faculty in Data Analysis & Visualization.
In partial fulfillment of the requirements for the degree of Master of Science.
Approved in September 2022.
Advisor: Eleanor Frymire.

Abstract

Robotics have been introduced to perform tasks traditionally done by humans.
They complement or substitute human labor, eliminating human involvement in:
- Hazardous environments
- Heavy lifting
- Toxic substances
- Repetitive low-level tasks
Robotics are intended to be more efficient and cost-effective, saving money, time, and labor.
Societal opposition exists due to fears of losing employment, wages, and purpose.
Previous studies report fears that robotics will progressively reduce employment and wages.
This project addresses the social impact of robotics on the labor and employment market via visualizations accessible to academic and professional audiences.
- It looks at the positive/negative impacts on employment and wages.
- It examines the progression/transition over time.
- It identifies where in the world by industry these robotic "industrial revolutions" are mostly taking effect,
- It seeks to understand what is driving the need for this industrial revolution and its effect on the current and future state of employment.

Acknowledgements

Thanks to Eleanor Frymire, Jason Nielson, and Matthew K. Gold for their patience, support, and guidance.
Thanks to the rest of the faculty in the Data Analysis and Visualization program for all their teachings.
Thanks to his wife Marjorie Espinal and Father Marino Espinal for their patience and support.
Special thanks to his son Kelvin Jr. Marino Espinal.

Digital Manifest

Capstone Whitepaper (PDF)
Web Hosted Files
- Project Website: https://kespinal83.github.io/Capstone_Final/index.html
- Archived version of project website (WARC file)
Code and other Deliverables
- Zip file containing the contents of the GitHub repository: https://github.com/kespinal83/CapstoneFinal/blob/main/CapstoneFinal.zip

List of Variables

Buckets: Data in buckets for visualization grouping
Cards: Heatmap individual boxes
centerPos: Center positioning
colors: Static colors
colorscale: Gradient color
datasetpicker: Specific dataset selector
data: Specific dataset selector
dayLabels: Static day of week labels
datasets: Datasets
days: Days of week selector
end_dist: End destination of arc
flyerAltitude: Arc altitude measured away from globe
graticule: Intersecting lines of latitude and longitude scale
gridSize: Grid size ratio
heatmapChart: Heat map variable for settings and attributes
height: Height of object
initRotation: Rotation speed and settings of globe
legend: Legend width and height variables
legendElementWidth: Legend element width settings
margin: Object margin settings
maxElevation: Maximum elevation away from globe settings
n_segments: Segments measurement
offsetX: Offset measurement from x axis
offsetY: Offset measurement from y axis
path: Calculated path on projection
projection: Projection of map type
radius: Radial object sizing
scaleExtent: Object size based on scale
sensitivity: Sensitivity of interaction and globe animations
skyprojection: Layer away from globe
start_dist: Start distribution of arc points
svg: SVG object in view
swoosh: Arc line
timeLabels: Static time of day labels
times: Times of day selector
width: Object width measurement

Glossary of Functions

convertToTimeStamp(): Timestamp conversion.
dash_offset(): Defines an offset on rendering the associated dash array.
dash_size(): Allows for geometric lines and points across the globe.
data(): Attaches data of any type to DOM elements.
dragged(): Allows for dragging and interaction of globe.
flying_arc(): Controls geometric points for arc.
get lineal(): Obtains line value.
locationalongarc(): Location overlap where arc exists
path_intersection(): Function for events that occur at path intersection.
position_labels(): Positioning of labels on the globe.
ready(): Used to make a function available after loading the document.
refresh(): Refreshes data and objects during actions and events.
refresh layers(): Refreshes flyers on rotation / interaction.
refreshLandmarks(): Refreshes landmarks on rotation / interaction.
tsvfile): Converts data from TSV file.
zoomed(): Function for zooming capability on the object.

Note on Technical Specifications

Solutions used: Visual Studio Code, GitHub Desktop, GitHub repositories.
A local directory for objects is created and made available to Visual Studio Code.
Dependencies (local or web-based) must be referenced in advance.
A repository is created on GitHub.com to host the files for storage and Webhosting.
GitHub Desktop is used to transfer and apply updates to files.
D3 version 5 is used for the globe's dynamic functionality.
The repository contains an HTML file (index.html) that serves as the main page structure.
CG.html / HM.html/ TL.html are all three main visualizations working independently, utilizing one main CSS file (style.css) and various JavaScript files.
Subfolders contain all dependencies for this project, including data files required by the visualizations.
To launch the project locally, install Live Server (v5.7.5 or higher), navigate to the root directory of the project files, right-click, and run on any of the .html files.

Introduction

An industrial robot is an automatically controlled, reprogrammable, multipurpose machine resembling a human that can replicate specific human movements and functions.
The topic of robotics in the workforce and labor sector is polarizing.
Robotics aims to eliminate human involvement in hazardous environments, heavy lifting, toxic substances, and repetitive low-level tasks.
They are meant to be more efficient and cost-effective.
Societal opposition has risen due to fears of losing employment, wages, and purpose.
Previous studies report fears that adopting robotics will progressively reduce employment and wages.
Robots' impact on the labor force varies throughout different industries, geographic areas, societies, and populations.
The effect is mainly in manufacturing industries.
The automotive industry employs 38% of existing robots (up to 7.5 per thousand workers).
Tesla deploys 75% of production lines automated by robotics vs. 25% humans.
The electronics industry employs about 15% of robots, while plastics and chemicals use 10%.
Labor workers in these industries tend to see the most dynamic turnover to robotics.
Adverse effects are estimated for workers in services, construction, etc.
The impact of robotics to human ratio in several industries was consistent across the board.
Robotics started as far back as the 1950s.
The first industrial robot was developed in 1954.
Mass production began in 1961 in a General Motors factory to automate die casting and handling spot welding.
Shortly after, fabrication of motorcycle frames and precision insertion tasks followed.
For every robot added per 1,000 workers in the U.S., wages declined by 0.42%.
The employment-to-population ratio went down by 0.2 percentage points (about 400,000 potential jobs lost).
Since the early 90s, the increase in robots (about one per thousand workers) reduced the average employment-to-population ratio in a zone by 0.39 percentage points, and average wages by 0.77%, compared to commuting zones with no exposure to robots.
Adding one robot to an area reduces employment by about six workers.
Robotics will mainly impact occupations where routine, repetitive tasks are daily.
Both sexes are affected, but males are impacted more than females in manufacturing and complex labor jobs.
Robotics primarily impacts workers without college degrees far more than those with a college degree.
Robot adoption does not positively affect workers with master's or advanced degrees.
Industrial robots do not directly complement high-skill workers, unlike other technology.

Project

The main goal of this project is to create awareness and expose a topic not commonly shared in the media.
Robotics comes with mixed feelings: "cool" automation with an underlying impact on the relationship between robot and laborer.
The project attempts to visualize answers to the following questions:
- What are the positive or negative impacts of introducing or utilizing robotics into the workplace on employment and wages?
- What has been the progression/transition over a time that presents the adoption and induction of robotics and the effects on the labor and employment market?
- Where in the world by industry are these robotic “industrial revolutions” mostly taking effect?
The project researches the topic and looks for key point indicators (KPI) to visualize in a form other than text to create a semi-narrative / exploratory visualization.

Capstone Visualization

The visualization aims to convey a complex story to an audience with varying knowledge, allow user interaction, and be modular for future updates.
The first point is to research and extract data from academic papers, articles, and readily available datasets.
Data drives what type of visual charts will make sense and aid design decisions.
The initial idea was to have an orthographic projection to convey data points across the world in an interactive format.
A second visualization involved an interactive timeline.
The third aimed to develop a custom merge of the histogram and heatmap to convey the effect of robotics on human workers by industry across different countries.
The first step was to draft a potential coherent page of all three visualizations and desired functionality.
The initial iteration (figure 1) utilized blank space and had a central orthographic projection.
A word cloud would display words originating from a specific country on hover.
The second visualization was a simple timeline, and the third was a heatmap showing intensity in robotics integration over time.
Functionality would allow end-users to interact by hovering over visuals and bringing them into focus.
The visualization design was changed to be clean and simple, with futuristic colors associated with cyberpunk neon-style or retro 80s colors.
Shades of blue-white, grey, and orange were used, with darker tones of the colors to set the precedence of mature visuals.
The page structure was designed to be symmetrical and evenly distributed.
The sections within the page must be symmetrical and evenly distributed.
Each line had different header styles to prioritize each title.
A word wrapper technique called spinny-words keeps "I am" static while cycling through a table of values that indicate common words describing human opinions and emotions found in the research papers and media reviews.
Flexbox technique for the border under the spinny-words technique to allow for color functionality below the title.
The first visualization is an orthographic globe representing countries where robotics has been embedded in the workforce.
end-users can interact with the world to find data points for countries.
every country would have data available and enable end-users to explore more.
On the hover of each country, orange will shade over and display minor metrics.
A timeline was added to help end-users understand the change and events that led up to the current state of robotics.
An unconventional design was created with no line in the center.
The size of each circle indicates the point in time and milestone relevancy in length on the hover of each historical event.
A time and label of the event will appear, and the circle will slightly animate to express focus on a point in time or event.
Selecting an event will open a text box.
The third visualization was a heatmap altered to represent the industry's time and intensity of robotics.
Three countries (China, Japan, and the United States of America) were focused on.
The color scheme is the same throughout, and animation bounce was added on change of country transition.
Each card in the heat map has rounded edges.

Course of Study Relationship

This project relates to Data Analysis, Data Studies, and Data Visualization.
For Data Analysis, the fundamentals of working with data to manage, develop, and systematically work up into a curated solution to describe, illustrate, condense, recap, and evaluate data.
Statistical analysis techniques learned were critical in reading research papers about embedding robotics and statistical examinations of labor and workforce.
Since data is not widely available, there was much cleanup and extraction from the text to devise and simplify the data.
Regarding the relationship to Data Studies, ethical thought processes of how technology such as robotics impacts social, political, and cultural aspects was applied.
The ethical problem that the introduction of robotics is causing the human workforce was presented with a practical yet neutral intent.
Challenge foundations and learned skills in data visualization techniques.
Skills learned to create a coherent functional visual that will help convey this topic and engage the user.
Engaging and effective information displays utilizing web-based technologies, including HTML, CSS, and D3.js, were created.

Evaluation

More extensive and intricate than originally thought.
Audiences can be captured, but there can certainly be improvements.
Limitations:
- Most existing research data is specific to the first-world represented societies and economic countries.
- Some second-world countries have no data on any third-world countries at all.
- Most research looks at major first-world countries such as the United States, China, Japan, etc.
- Regardless of where robotics integrated itself into the industry, it became more “mainstream" where applicable after the late 80s and early 90s.
Coding setbacks due to a lack of knowledge and resource limitations.
- A rich animation visualization was desired, like the globe flying in.
- Nations would increase in size on hover or click and go back to scaled size.
- Functions would fail when compounded.
- All three visualizations are running independently and combined into the main page by iframe containers.
- Running all three on one page slows down the page.
Style decisions were challenging for the timeline visualization.
The third visualization was just a matter of how much data was wanted to display.
State management was one function that failed.
- If Japan is selected, all other visualizations will update to data from Japan.
The project succeeds in conveying the topic to a general audience.
Visualizations are exciting and engaging.

Project Continuity

Collect more data.
- Society would have to survey and develop more data and studies around this topic, or, if data exists, make it more available.
- Build a python script that would go out to the web and look for data related to this topic.
Create a better file structure and coding.
- Fewer files, be more embedded, and be less resource intensive.
- Address performance issues.
Have more fluid narrative interactivity such as state management.
- As the end-user scrolls throughout the page, visualizations would automatically and cohesively auto-update.
Overshoot and then tone down the experience due to restrictions, knowledge, or just plain it does not work as imagined.
- For instance, the globe to fly over the page into position and then slowly rotate on the axis.
- The ability to interchange between visualizations in a carousel-type animation.
Programming languages outside d3.js may be a better fit for some resource-intensive visualizations and animations.
Consideration of structure and hierarchy of side and visualization components.
Improvements to where data is versus visualizations also where symmetry works vs. not.
The exploratory visualization can help educate on this topic for multiple audiences, increasing awareness.

BIBLIOGRAPHY

Acemoglu, Daron, and Pascual Restrepo. "Robots and Jobs: Evidence from US Labor Markets." 2017, https://doi.org/10.3386/w23285.
Acemoglu, Daron, et al. "Competing with Robots: Firm-Level Evidence from France." 2020, https://doi.org/10.3386/w26738.
Brown, Sara. "A New Study Measures the Actual Impact of Robots on Jobs. It's Significant." MIT Sloan, 29 July 2020, https://mitsloan.mit.edu/ideas-made-to-matter/a-new-study-measures-actual-impact-robots-jobs-its-significant#:~:text=The%20researchers%20found%20that%20for,loss%20of%20about%20400%2C000%20jobs.
DAĞLI, İbrahim. “Will Workers Be Unemployed Because of Robots? A Meta-Analysis on Technology and Employment.” Sosyoekonomi, 2021, https://doi.org/10.17233/sosyoekonomi.2021.04.22.
Dekle, Robert. "Robots and Industrial Labor: Evidence from Japan." SSRN Electronic Journal, 2020, https://doi.org/10.2139/ssrn.3670356.
Dixon, Jay, et al. “The Employment Consequences of Robots: Firm-Level Evidence." SSRN Electronic Journal, 2019, https://doi.org/10.2139/ssrn.3422581.
Dottori, Davide. "Robots and Employment: Evidence from Italy.” SSRN Electronic Journal, 2020, https://doi.org/10.2139/ssrn.3680743.
Morikawa, Masayuki. “Firms' Expectations about the Impact of AI and Robotics: Evidence from a Survey." Economic Inquiry, vol. 55, no. 2, 2016, pp. 1054–1063., https://doi.org/10.1111/ecin.12412.
McGaughey, Ewan. "Will Robots Automate Your Job Away? Full Employment, Basic Income, and Economic Democracy." 2019, https://doi.org/10.31228/osf.io/udbj8.
"Occupations by State and Likelihood of Automation - Dataset by WNEDDS." Data.world, 23 June 2017, https://data.world/wnedds/occupations-by-state-and-likelihood-of-automation.
Office, U.S. Government Accountability. "Workforce Automation: Better Data Needed to Assess and Plan for Effects of Advanced Technologies on Jobs." Workforce Automation: Better Data Needed to Assess and Plan for Effects of Advanced Technologies on Jobs | U.S. GAO, 23 May 2019, https://www.gao.gov/products/gao-19-257.
Tang, Chengjian, et al. “Robots and Skill-Biased Development in Employment Structure: Evidence from China." Economics Letters, vol. 205, 2021, p. 109960., https://doi.org/10.1016/j.econlet.2021.109960.