Abstract

This project looks at how COVID-19 affected jobs worldwide. The time period we examined employment rates was three years before the pandemic, the year that COVID-19 caused the shutdown (2020), and three years after the pandemic. Our findings show significant changes in employment, which highlights the pandemic's significant impact on the workforce. We also looked at the daily increase in COVID-19 cases, revealing the interconnected challenges societies faced during this unique time.

Introduction

In 2020, it was the year when the world slowed down due to the pandemic; businesses began to shut down one by one, which caused people to live without jobs. With no sources of income, the global citizens started to experience hardships in living: no money to pay for rent, no food to put on the table for their families, and no way to support themselves. Therefore, many companies, mainly in technology fields, changed their working mode from in-person to remote by using applications, such as Zoom and Google Meet, to ensure the company can operate with enough employees. Nevertheless, there were still a massively large number of unemployed people. 

As we know, COVID-19 is a contagious virus that people can easily be infected by close contact with an individual. The flu-like symptoms such as fever, runny nose, sore throat, coughs, and loss of taste and smell are common to infectors with high fatality rates due to the dismissal of vaccine until May 1st, 2020. In America, when the first confirmed case of COVID-19 was founded in Snohomish County, Washington, the fact brought panic to the society and later leads the lockdown of the country. Therefore, with the cases exponentially increasing, everyone started to live under strict regulations -- nobody was able to leave their home until further notice by the government. The global impact of COVID-19 has introduced unprecedented challenges, particularly in employment. The issue of getting a job has sparked widespread interest in understanding the profound economic and societal consequences. While these sources shed light on the significant disruptions to careers and the surge in daily cases during the pandemic, the employment trends among the three periods remain difficult to understand: the three years before the pandemic, the year COVID-19 caused the shutdown (2020), and three years after the pandemic.

As a result, the questions we aim to answer in this project are: 
- How were the employment rates before COVID-19 caused the shutdown in 2020?
- What were the immediate effects of the COVID-19 shutdown on employment in 2020?
- What changes have occurred in the employment landscape during the three years that followed the shutdown?
- Is there a relationship between COVID-19 cases in a year (either increasing or declining since 2020) and employment trends?
Our report provides insights into how the COVID-19 shutdown negatively impacted employment. When the shutdown happened in 2020, millions lost their jobs because of this virus. While inflation was on the rise, simple things like foods were getting very expensive. Having no job means not having any money to buy necessities. Where we currently are in 2024, unemployment still has not returned to its average before the lockdown. Although COVID-19 has slowed down and there are fewer cases due to vaccinations, we don’t want another relapse and millions of people to lose their jobs.

Data

According to our topic, this project aims to find the relationship between COVID-19 and employment status in America. Therefore, we access tables containing population data for the pandemic and labor force from official organizations. While the COVID-19 data set is from the World Health Organization and the working status dataset is from the Bureau of Census for the Bureau of Labor Statistics, these data include high reliability because they are collected by governors through surveys and reported statistics. Furthermore, both datasets are obtained from official websites so that the given information is ethical without any bias. Also, because the target audience for both datasets is American citizens, they are open to people who are interested in related topics. Therefore, both datasets can be used based on research purposes or personal use. When we get the datasets, the given tables are easy to understand but the format is unavailable for the R programming to read -- we sorted the xlsx file into a CSV file. When processing the organizing in datasets, we found that none of the values were missing, which indicates the cleanliness of both tables.
##   year total_employed_lf total_unemployed_lf non_lf
## 1 2017            153337                6982  94759
## 2 2018            155761                6314  95716
## 3 2019            157538                6001  95636
## 4 2020            147795               12947  99587
## 5 2021            152581                8623 100241
## 6 2022            158291                5996  99686
## 7 2023            161037                6080  99826
##    Date_reported Country_code     Country WHO_region New_cases Cumulative_cases
## 1     2020-01-05           AF Afghanistan       EMRO        NA                0
## 2     2020-01-12           AF Afghanistan       EMRO        NA                0
## 3     2020-01-19           AF Afghanistan       EMRO        NA                0
## 4     2020-01-26           AF Afghanistan       EMRO        NA                0
## 5     2020-02-02           AF Afghanistan       EMRO        NA                0
## 6     2020-02-09           AF Afghanistan       EMRO        NA                0
## 7     2020-02-16           AF Afghanistan       EMRO        NA                0
## 8     2020-02-23           AF Afghanistan       EMRO        NA                0
## 9     2020-03-01           AF Afghanistan       EMRO         1                1
## 10    2020-03-08           AF Afghanistan       EMRO        NA                1
##    New_deaths Cumulative_deaths
## 1          NA                 0
## 2          NA                 0
## 3          NA                 0
## 4          NA                 0
## 5          NA                 0
## 6          NA                 0
## 7          NA                 0
## 8          NA                 0
## 9          NA                 0
## 10         NA                 0
## The CPS dataset include 4 columns and 71 rows.
## The WHO COVID dataset include 8 columns and 51360 rows.
The first dataset is the CPS dataset, also known as the Current Population Survey. In the dataset, it includes the monthly survey of households over time. Therefore, each row of the data provides the labor force, employment, unemployment, and persons not in the labor force based on diverse demographic and labor force characteristics. In our project, we include the total employed labor force, the total unemployed labor force, and the total not in the labor force.

The second dataset is the WHO COVID-19 data. In the dataset, it includes the numbers of confirmed COVID-19 cases and deaths through official communications under the International Health Regulations. Therefore, each row of the data provides the new and cumulative confirmed cases and the new and cumulative death cases for different countries every seven days, starting from January 5th, 2020. In our project, we include the new confirmed cases and new death cases only in America.
  1. Current Population Survey(CPS)
  2. Global COVID-19 Data
##          date cases ave_case_rate deaths ave_death_rate
## 1  2020-01-05     0         below      0          below
## 2  2020-01-12     0         below      0          below
## 3  2020-01-19     0         below      0          below
## 4  2020-01-26     7         below      0          below
## 5  2020-02-02     2         below      0          below
## 6  2020-02-09    11         below      0          below
## 7  2020-02-16     4         below      0          below
## 8  2020-02-23    16         below      0          below
## 9  2020-03-01    33         below      1          below
## 10 2020-03-08   391         below     18          below
##   year total_employed_lf total_unemployed_lf non_lf total_lf
## 1 2017            153337                6982  94759   160319
## 2 2018            155761                6314  95716   162075
## 3 2019            157538                6001  95636   163539
## 4 2020            147795               12947  99587   160742
## 5 2021            152581                8623 100241   161204
## 6 2022            158291                5996  99686   164287
## 7 2023            161037                6080  99826   167117
After loading the datasets, we clean up the table to generate into dataframes that can make the analysis process easier. We have add in a new categorical column to the COVID-19 dataset to show whether if the confirmed and death cases in America is above or below the global average. On the other hand, we also creates a new numercal column to the CPS dataset to introduce the total working population over time.
##   year total_employed_lf total_unemployed_lf non_lf total_lf total_case
## 1 2017            153337                6982  94759   160319          0
## 2 2018            155761                6314  95716   162075          0
## 3 2019            157538                6001  95636   163539          0
## 4 2020            147795               12947  99587   160742   18890446
## 5 2021            152581                8623 100241   161204   32988414
## 6 2022            158291                5996  99686   164287   47140633
## 7 2023            161037                6080  99826   167117    4417336
##   total_deaths
## 1            0
## 2            0
## 3            0
## 4       342920
## 5       469667
## 6       267389
## 7        83835
After cleaning up the datasets, we use the sources to create a new dataframe that combines both datasets. In the merged dataframe, we use "year" as the merge key to combine the values we have for CPS and COVID-19 datasets.

Methods

Data Selection: We carefully chose relevant variables from both datasets, focusing on COVID-19 cases, deaths, and labor force statistics such as total employed, unemployed, and non-labor force individuals. We select these variables to analyze the impact of the COVID-19 pandemic on employment trends and vice versa.

Data Processing: Using R programming, we include the "dplyr" package and "tidyverse" package to process the data by filtering observations from 2017 onwards to concentrate on recent trends. Additionally, we rename and select specific variables for clarity and consistency across datasets to ensure the integrity of our analysis.

Data Merging: We join the datasets using "year" as the merge key to align COVID-19 and labor force statistics chronologically. The new data frame allows for a comprehensive analysis of the relationship between public health and economic indicators over time.

Handling Missing Values: To maintain data integrity and avoid bias in our analysis, we address missing values by replacing them with zeros for COVID-19 cases and deaths.

Statistical Analysis: Our statistical analysis primarily involves calculating averages for COVID-19 cases and deaths to provide summary insights. While more complex statistical methods, such as linear regression, are not accounted for in this analysis, we conduct descriptive and exploratory data analysis to uncover trends and patterns.

Sample Selection: The sample selection focuses on observations from the United States of America, providing insights into the COVID-19 pandemic's impact on a specific country. The decision is based on the availability of reliable and comprehensive data for analysis.

Visualizations: Visualizations are created with the `ggplot2` package in R programming to communicate trends and insights derived from the data. We make bar plots and line plots to represent COVID-19 cases, deaths, and employment trends over time clearly and intuitively.

Results

The bar plot showcases the trend of confirmed COVID-19 cases over time in the United States. We observe significant fluctuations in the number of cases, reflecting the evolving nature of the pandemic.

Similarly, this bar plot demonstrates the progression of COVID-19 death cases over time in the United States. The plot reveals the grim reality of the pandemic's toll on human lives.

This plot illustrates the status of COVID-19 confirmed cases over time, categorizing them as above or below the average case rate. It provides insights into the severity and fluctuations of the pandemic's impact on case numbers.

This plot depicts the status of COVID-19 death rates over time, categorizing them as above or below the average death rate. It helps in understanding the mortality trends associated with the pandemic.

The line plot presents the working status in America, distinguishing between employed and unemployed populations from 2017 to 2023. We observe fluctuations in both employment and unemployment levels, with a notable spike in unemployment following the onset of the pandemic.

This scatter plot displays the relationship between total employment and total unemployment in the United States over time. A clear positive correlation is observed, indicating that as employment increases, unemployment increases.

The line plot illustrates the relationship between total COVID-19 cases and total unemployment in the United States from 2017 to 2023. We observe a general trend when the number of COVID-19 cases increases, unemployment tends to rise. However, this relationship is complex and may be influenced by various factors.

In contrast to the previous plot, this one shows the relationship between total COVID-19 cases and total employment in the United States from 2017 to 2023. Interestingly, we see that as COVID-19 cases increase, employment levels decrease.
The analysis reveals intricate relationships between COVID-19 dynamics, economic indicators, and societal impacts. While there's a general association between increasing COVID-19 cases and rising unemployment, nuanced patterns emerge when considering employment levels. These insights can inform policy-making and public health strategies to navigate and mitigate the effects of the ongoing pandemic.

Discussion

The results highlight the following founds regarding unemployment: the number in the U.S. was lower before the shutdown, the number spiked in the year of the shutdown, and the number has not returned to its average before the shutdown three years after 2020. These insights tell us that the COVID-19 shutdown in 2020 hurt unemployment in the U.S. and continues to impact current employment rates three years later. The outcomes should also be noted that there is no correlation between COVID-19 cases in a year and employment trends -- even though the total COVID-19-related cases kept increasing during the two years after the shutdown, the unemployment rate started to decrease in those two years.
The negative impact of the shutdown is evident through the following understandings. The COVID-19 shutdown required remote work for many people, but the issue is that not all work is suitable for online. Except for healthcare workers and a few other essential personnel, people whose work could not be achieved online would not have a job due to the result of the shutdown. On the other hand, the changes in supply and demand also hit society hard. People were not going out as often as before, and this caused many businesses to shut down due to the lack of customers. Regardless of which route created barriers, it does not change the fact that the shutdown seriously impacted the unemployment rates for the worse. However, without further research, there might be extraneous factors that could also contribute to the low unemployment rates. For instance, 2020 was an election year in America, and the inauguration of a new president could also be contributing to why average unemployment has not returned to pre-shutdown averages -- this is a significant limitation in our conclusion. To support the results we have drawn, we would have to see if and how much unemployment fluctuates on average in any given election year. Even then, we can see the possible fluctuations in years past, but that data would not likely yield valuable information for coming years because years after the shutdown should also be considered. As such, future work would require data collection on unemployment during the following election years. Only after this could we further analyze a potential fluctuation in unemployment during presidential election years and determine if it was an influencing factor in our current data, or if the COVID-19 shutdown was the driving force behind unemployment spiking in 2020.

Conclusion

In 2020, the world experienced a shift as the COVID-19 pandemic led to widespread business closures and job losses. Companies started to adapt to remote work, but this still left a substantial portion of unemployed people. Our project goes into the employment landscape across three different periods: three years before the pandemic, the year when COVID-19 caused the shutdown (2020), and three years after the pandemic. The questions guiding our investigation aim to explore the relationship between COVID-19 cases and employment shifts. Our findings reveal a negative impact on employment during the 2020 shutdown, with many people losing their jobs. To conclude, our report shows the consequences that the COVID-19 lockdown had on employment status. Despite the progress and decreased COVID-19 cases due to vaccinations, unemployment rates still have not recovered to normal levels. COVID-19 still exists, and different variants appear continuously. Knowing what the original virus did to employment rates in 2020, we want to prevent another relapse where millions of people will lose their jobs.