Name

Affiliation

Course

Lecturer

Due Date

Abstract

The world has experienced a technological storm in recent times. Technological growth and innovations have been welcomed by most nations as they increase efficiency and production levels, especially in the manufacturing sector. In recent times, technological innovations have gone ahead to develop Artificial Intelligence with the most recent innovation being the Lucid AI. The development of AI has arisen a heated debate between techno-optimists and techno pessimists. The techno pessimist believes that the growth of technology and especially artificial intelligence would lead to future humankind unemployment levels increase and low wages in what is termed as “technical substitution”. On the other hand, techno-optimists believe that the growth n technology is likely to lead to an economic expansion and thus lead to the creation of new roles and jobs thereby reducing unemployment rates. This research seeks to contribute to this debate by analyzing three of the most common AI tools in the manufacturing sector using a case study of the UK.

Keywords: Artificial Intelligence, innovations, Lucid AI, Techno optimists, techno pessimists, technical substitution.

Table of Contents

Abstract 1

Chapter 1: Introduction 1

Background 1

Statement of problem 2

Objectives 2

General objectives 2

Specific Objectives 3

Research Questions 3

Scope 3

2.0 Literature review 3

2.1 Job Polarization 3

2.2 Artificial intelligence 4

2.2.1 Robots 5

2.2.2 Automation 6

2.2.3 Machine learning 8

3.0 Methodology 8

Introduction 8

Research Design 8

Empirical Model 8

Sampling Criteria 9

Data collection methods 9

Data Analysis 9

Measurement of Variables 10

Multicollinearity 11

Autocorrelation 12

Heteroscedasticity 12

Stationarity Test 12

Chapter 4: Empirical results 12

Descriptive statistics 12

Correlation matrix 13

Diagnostic tests 13

Heteroscedasticity 13

Autocorrelation 13

Multicollinearity 14

CHAPTER 5: CONCLUSION AND POLICY IMPLICATION 15

Conclusion 15

Policy recommendations 15

References 16

Appendix 18

Effects of Manufacturing Technology on Human Labour in the UK.

Chapter 1: Introduction

1.1 Background

In 1929, Keynes theorized that rapid spread in automation technology would bring “technological unemployment” (Keynes 1931). Technology is the application of scientific knowledge into practical works (Mishra, 2009, p. 187). In recent times, the world has experienced quite a tremendous shift in technology as innovations continually continue to shape most industries. The manufacturing sector has been the most affected as recent innovations continually increase effectiveness and efficiency in the manufacturing sector. One of the most recent technological innovations in the development of artificial intelligence. Artificial intelligence refers to the simulation of human intelligence in machines that are programmed to think like humans and mimic their actions (Simmons & Chappell, 1988, p. 42). The manufacturing sector is one of the beneficiaries of technological innovations. The manufacturing sector has benefited from the automation of processes and thus increased effectiveness in the production process. However, the automation process has come along with some serious dangers as managers substitute human laborers for machines. Most manufacturing private companies have a primary goal to increase sales revenues through production, subsequently, most of these managers have substituted human labor for machines that have exhibited more effectiveness. The dynamics of the labor market have immensely changed as numerous individuals continue to be displaced; by machines. The trend is quite worrying and thus researchers have sought to establish the real cause for these changes in the labor market.

Technology has immensely revolutionized the manufacturing sector easing the production process. Over the years, technological innovations have continually increased and so has the rate of unemployment. In addition, wages have continually decreased as employees are forced to take wage cuts as managers attempt to substitute human labor for machines. Researchers have referred to the substitution of human labor for machines as the displacement effect as most firms shift towards improving their productivity through automation of basic predictable processes (Nawaz Sharif & Kabir, 1976, p. 356). McKinsey Global Institute conducted an empirical study and forecasted that 4 million Americans would lose jobs by the year 2030 which accounts for a quarter of the total workforce (Smit et al., 2020).

Recent times have seen rapid growth in technological innovations in the manufacturing sector as issues about artificial intelligence take precedence. Issues on the future of the human workforce have become a major concern as machines have proven to be more effective especially in performing predictable tasks (Zhang et al., 2017).

The matters of technological innovations have been a key issue requiring the intervention of all key stakeholders. The issue of innovative technological progress has a key impact on employees who could be rendered jobless soon. Researchers estimate that 70 machines can perform predictable tasks that could previously be conducted by 10,000 human workers (Zhao, 2018). In addition, employers could be in a dilemma whether to shift to more productive labor through machines or maintain the current human labor. Efficiency is one of the key considerations most managers focus on and this could see most of them displace human labor for robots and machines (Zhao, 2018). Due to the serious adverse effect technology could cause on the workforce, there is a dire need for the government to intervene and ensure that employees don’t lose their jobs to machines as manufacturers try to increase efficiency.

Despite the evidence of job polarization and forecast, on technological unemployment, there is still little research conducted towards a determination of the impact of job polarization in the manufacturing sector.

1.2 Statement of the problem

The world has experienced tremendous technological innovations, especially in the manufacturing sector. However, researchers have greatly criticized the recent technological trends such as AI as having a displacement effect and leading to technological substitution as theorized by Keynes (Keynes 1931). There has been heated debate among scholars as they seek to establish the impact of technology on employment status. This study focuses on the three major types of Artificial intelligence technologies in the manufacturing sector and their overall effect on the unemployment status and trends in the UK.

1.3 Objectives

1.3.1 General objectives

This article seeks to contribute to the heated debate on the effects of technology on unemployment by establishing the core effect of technological advancements in the manufacturing labor market.

1.3.2 Specific Objectives

1. Determine whether systems automation is a significant determiner of unemployment,

2. Determine whether machine learning is a significant determiner of unemployment, and

3. Determine whether the increase in robots is a significant determiner of unemployment in the UK manufacturing sector.

1.4 Research Questions

1. What is the effect of automation on the unemployment status in the UK manufacturing sector?

2. What is the effect of increased robots on the unemployment levels in the UK manufacturing sector?

3. What is the effect of increased machine learning on the unemployment levels in the Uk manufacturing sector?

1.5 Scope

This research is aimed to establish whether there is a significant relationship between unemployment rates in the UK and technological innovations change in the manufacturing sector over the years 1990-2020. Finally, this study will propose policies to remedy the situation based on the findings. This research seeks to establish whether there has been a significant relationship between unemployment levels in the UK and technological innovations such as automation, machine learning, and artificial intelligence

2.0 Literature Review

Technology is quite a fundamental aspect of growth and sustainability in the economy. However, a rapid increase in technology is perceived to be dangerous and hurt employment. Back in 1929, John Maynard Keynes theorized that rapid growth in technology and specifically automation would in the future lead to what he termed as “technological unemployment” (Keynes 1931). Wassily Leontief is quoted in Curtis (1983,8) having prophesied that, “Labor will become less and less important…. More and more workers will be replaced by machines. I do not see that new industries can employ everybody who wants a job”.

Recent studies have also sought to establish the likelihood of substitution of human labor. Based on empirical findings, robots and artificial intelligence developments are greatly linked to massive job losses (Vermeulen, et al, 2018; Hall, & Kramakz, 1998).

2.1 Job Polarization

According to Goos and Manning (2007), job polarization is a situation where middle-skills jobs seem to disappear relative to the lower jobs requiring low skills and the top-level jobs requiring high skills. The most recent research conducted by Salvatori et al., (2018 sought to establish the extent to which job polarization has affected the UK market. Based on Salvatori et al., 2018, technological advancements are one of the major causes of job polarization in the UK market. It is much easier to program machines to perform routine tasks most of which are those requiring medium skill levels. Based on the findings, there is a significant effect of skills level in job polarization. Job polarization mainly affects the middle-skilled workers who perform routine jobs that machines could be codified to perform (Goos, Manning, and Salomon’s 2014).

According to Feng and Graetz 2020, training requirements are one of the most significant factors that contribute to job polarization. According to empirical research conducted by Feng and Graetz 2020 on the UK market from the year 1980-2008, there is a high correlation between job polarization and skills level required (Feng & Graetz, 2020, p. 2250). When firms are faced with situations where two situations requiring further training, firms tend to automate the job requiring more training. Training requirements and complexity of tasks account for quite a large percentage of the US job polarization. The study discovered that recent technological advancements have had quite numerous effects on the human labor market as firms seek to establish what processes to automate. Based on the research considering data from 1980-2008, firms mostly automate processes that are routine and which seem complex to undertake for workers. In addition, firms seek to evade high training costs and tend to automate such tasks that require the high training of employees (Feng & Graetz, 2020).

A study conducted on 12 European countries has discovered that there is actual evidence of job polarization (Peugny 2019). Based on the European Union Labour Force Survey (EU-LFS) there has been a continued decline in middle-skilled jobs and thus evidencing job polarization. Based on empirical shreds of evidence, there has been a massive increase in high-skilled jobs and low-skilled jobs whereas the middle-skilled jobs have continually declined over the years. Researchers fear that this may continually increase the inequalities gap and seriously compromise the Sustainable Development Goals (SDGs) towards equality in wealth and income. The research conducted by Peugny (2019) over 20 years it is quite evident that middle-skilled jobs are slowly disappearing. The trends in job polarization have been linked to technological advancements which tend to replace most of the routine, middle-skilled jobs.

A research conducted on the UK market by Goos et al., 2014, it is quite evident that the UK market is moving towards job polarization. The employment structure is becoming more and more polarized as technology continues to revolutionize the world. Goos et al. conducted in-depth research building onto previous 2009 and 2010 research intending to establish the extent of job polarization on the UK market. Based on empirical findings, job polarization is quite evident from the past three decades. Technological advancements are one of the primary factors that have immensely contributed to job polarization in the UK market. With the increased level of technology, most of the middle-skilled jobs are substituted by machines in what researchers have termed as the displacement effect.

2.2 Artificial intelligence

Most recent innovations have seen the development of artificial intelligence in robots manufacturing. According to (Williams, 2020) artificial intelligence is the simulation of human intelligence in programmed machines enabling them to reason and perform tasks just like humans. Artificial intelligence has been well embraced in the manufacturing sector having been associated with unlocking insights previously unattainable. With artificial intelligence, the manufacturing sector can increase its production capabilities and thereby meet the everchanging customer expectations (Salehi & Burgueño, 2018, p. 188). Recent studies have shown that the manufacturing sector remains the top beneficiary of the artificial intelligence process. Through AI, managers can now reduce unplanned downtime and now increase production standards and quantity (Piovesan & Kozman). Artificial intelligence despite leading to effectiveness has been a huge blow to human labor. Artificial intelligence means that machines can perform the basic functions of a human being more effectively and with precision. In the long run, machines are quite cheaper as compared to human labor, as a result, the future for human labor seems dull as more managers may tend to replace human labor for machines. A study conducted by (Ford, 2013, p. 39; King et al., 2017) shows that artificial intelligence is the future of the manufacturing sector as most routine jobs may end up being performed by machines.

2.2.1 Robots

One of the most recent scientific accomplishments was the development of robots. Robots are machines that are usually programmable and remotely controlled (Rollinson et al., 2012, p. 2046). One of the most unique features distinguishing robots from other machines is their ability to mimic human behavior and perform complex activities with greater precision and accuracy. The development of robots has been greatly criticized as it puts to risk the future of human labor. Despite the numerous criticisms, most nations have embraced such technological advancements arguing that these innovations would be crucial to their future effectiveness in most sectors.

According to Acemoglu and Restrepo, robot development is quite crucial for economic growth and development (2016). However, the WEF (World Economic Forum), robots will handle 52% of the tasks currently performed by humans. This is close to twice the current tasks performed. The seismic sharp increase in robot operations is also likely to see the development of new roles as humans will have to increase their skills to keep up with the revolution by robots. The swiss international firm said that today, robots only perform 29% of the jobs performed by humans, however, this is likely to change as, by the year 2025, robots will approximately perform 52% of the jobs performed by humans (Acemoglu, D., & Restrepo, 2016). The Phys organ study conducted in 2016 however maintains that the simultaneous change in machines and algorithms could see an increase in 133 million jobs in line with the 75 million jobs that will be lost by the year 2022. The study maintains that most sectors will be affected and especially the industrial sector where tasks are performed based on routine. Non-routine jobs requiring advanced human interactions such as creativity may however be less affected and could experience increased demand as more displaced individuals shift towards such fields.

The most recent research was conducted by Rodgers and Freeman (2019) on the actual impact of robots on the post the great recession _from 2009-2017. According to the article, this period exhibited a steady job growth and a fairly stable economy, it was also in this time that the growth of robots more than doubled. Despite there being an immense increase in robots, there is still limited research linking their growth to an increase in unemployment. The article also establishes that the growth of robots has a minimal effect on national unemployment since displaced workers easily find other jobs especially in a fairly stable economy with low unemployment. However, the article acknowledges that the displacement effect of robots leads to a decline in the total wage compensation received by workers. According to Rodger and Freeman (2019), the displacement from robots is quite possible but with previous booms, the economic expansion was so great and thus offsetting any job losses. The main evident factor is that robots tend to displace workers in specific groups while other sectors remain fairly unchanged. According to the findings on the impacts of robots in the rust belt(Middle east), robots have a significant effect on some groups of workers (young and less educated). Increased uptake of robots in the manufacturing sector is likely to lead to a decline in wages as the article estimates that an increase in 1 robot per thousand workers is associated with a 4 – 5% decline in wages.

Recent studies have painted quite an ugly picture depicting robots as the major cause of job loss. However, researchers such as Surrowecki (n.d) and Fleming (2018), do not agree with the numerous findings and believe the predictions made are misleading. Based on the scholarly finding, there has been a massive development of robots through the unemployment status was quite low before the Covid 19 pandemic. As a result, most states in the US were facing labor shortages as the robots continually increased. This is a clear indication that the market is performing below its capacity. The development of robots is meant to complement human tasks and not replace them. The study also focuses on the fact there has been quite limited research on the effects of robots on the actual unemployment rate. According to Suroweicki (n.d), pieces of evidence also show a low rate of job churn as it would be expected in a situation where robots were remaking the labor market. The author however concludes by acknowledging that automation and artificial intelligence affects the labor market bat believe the prior forecasts are farfetched and exaggerated.

In another research conducted by Peter Fleming, the researchers’ predictions that robots may want to displace human labor is strongly disputed (Fleming 2018). According to research, there is a huge debate between techno pessimists and techno-optimists on the future of technology on future employment, especially in human labor. According to Fleming, in previous technological advancements including computerization and machine learnings, there have been quite low levels of unemployment and displacement of human labor. Using the same argument, Fleming believes that there will be little or almost no effect of artificial intelligence and robots on the employment status soon as techno pessimists have predicted

In the near future, approximately half of the current jobs in the UK and United States could be automated, the McKinsey Global Institute (2017) says. 57 percent of jobs in the OECD are susceptible to mechanization within the next two decades, 69 percent in India and 77 percent in China in a related study of Oxford Martin School (2016). In contrast, highly cognitive jobs are now vulnerable to automation methods that focus on repetitive manual work. With a rise in artificial intelligence, the then too human manual jobs such as hairdressing are now at risk of being automated and replaced by Robots

2.2.2 Automation

In February 2021, McKinsey Global institute forecast that 45 million which is close to one-quarter of the population would lose their jobs due to automation by the year 2030 (Sevakula et al., 2020). The 2021 estimate was an increase from its 2017 estimate where it predicted that only 39 million would lose jobs from automation. Historically, most firms tend to replace the workers fired during recessions with machines.

According to Goldberg, (2012, p. 2) automation is a process of application of technologies to produce with minimal human intervention. Technological innovations have been continually increasing in recent times. As technology rapidly increases, the manufacturing sector has faced a tremendous shift as more and more operations are automated. Researchers are quite concerned with increased automation processes since more and more previously routine jobs are being taken up by machines. Automation has a long history in humankind history. However, the recent automation processes as a result of artificial intelligence have greatly put to risk the future of human labor. The recent innovations seem to be shifting from the conventional slow and less effective human species to a more sophisticated system of machines that can perform quite numerous tasks and with so much ease (Stone et al. 2016).

It is quite evident that the automation processes innovations and technological advancements will lead to job loss in the near future. Parschau and Hauge sought to establish the effect of automation in the apparel industry to ascertain the extent to which automation processes have on the manufacturing sector (Parschau & Hauge, 2020, p. 127). According to the study, they discovered that based on previous years’ data, there is no significant relationship between unemployment rates and automation processes in the manufacturing sector. The study also discovered that in some instances, the automation processes may lead to an economic expansion and creation of more positions thereby decreasing the unemployment levels (Parschau & Hauge, 2020, p. 130).

Managers tend to shift to more effective production means that will increase their output levels and minimize their cost of production. Automation is one of the best alternatives towards reducing production costs and attaining effectiveness. Bhattacharyya conducted a study to ascertain the Corporate Socially Responsible (CSR) managers’ justification towards adoption of automation processes and its effect on job loss (2008, 97). The findings indicated that the CSR managers rationalized the adoption of automation technologies from a PPM (push-pull Mooring) perspective from a firm-centric viewpoint (Bhattacharya et al., 2008, p. 98). While system thinking with a fair-market ideology rather than a normative justification was used for justifying from a community (social) center perspective, utilitarian thinking was applied instead of reputational egoism (Bhattacharyya, 2008). The study perceives it rational that the CSR managers undertake community activities in an attempt to caution displaced individuals from poverty. It is quite evident that there is some level of job loss emanating from automation, however, the study emphasizes that CSR managers also have to reduce this displacement effect (Bhattacharya et al., 2008, 100).

“Make in India” is one of the recent developments that had Indians optimistic of a decrease in the unemployment status (Green 2014). However, the recent developments have greatly affected these optimistic statuses as the technological automation process and limited capital greatly compromised employment achievements. The automation process is perceived to have a deleterious effect on the creation of employment in different sectors of the economy manufacturing sector included (Coupe 2019 p 1290). According to Green (2014), there is a range of technologies that are involved in industrial processes. These technologies vary greatly especially in their effects on employment. Some of the technologies are perceived to lead to an increase in the employment status whereas other technologies compromise the employment status through the displacement effect.

A recent study was conducted on small and medium metalworking firms in Ohio. The study sought to establish the perception of automation on the firms’ owners and the overall effect new automation processes have on workers. The main aim of the study is to establish whether new technologies bring about redeployment or robocalypse (Waldman-Brown, 2020, p 101). Based on the findings, automation in the manufacturing sector brings about efficiency in the production process with minimal displacement effect. Automation processes are mostly associated with the deployment of workers as they further their skills to take over different roles within the same firm. The automation process is mainly associated with the expansion of firms rather than the displacement of workers from their jobs (Waldman-Brown, 2020, p. 102).

2.2.3 Machine learning

Machine learning was one of the earliest technological innovations and advancements in the manufacturing sector. Machine learning is a method of data analysis that automates analytical model building (Jordan & Mitchell, 2015, p. 258). Machine learning plays a crucial role in the technological innovations process since it makes it easier for simulations and thus easing an understanding of the functionality of machines. According to (El Naqa & Murphy, 2015, p. 4) machine learning is a method of artificial intelligence where machines could identify patterns from a given data set, analyze and make decisions with minimal human intervention. The previous technologies still require human intervention to run them and make critical decisions. However, recent software automation and machine learning have seriously questioned the future of human employment. In a scenario where machines could have analytical skills and make a judgment with greater precision than humans, it is then logical that most firms may in the future opt to replace these workers with machines that can make better decisions (Boselli et al., 2017, p. 338). Lucid AI is one of the latest and most sophisticated artificial intelligence tools that immensely mimic human behavior (Neubert & Montañez, 2020, p. 198). The new technology comes equipped with the speed, power, and scalability of modern computing.

With the rapid increase in technological advancements, the entire world is at risk of “technological unemployment. This phenomenon has been theorized since the times of Keynes and with the recent machine learning and artificial intelligence, scholars still exhibit skepticism on the future of human labor. Despite these trends, Wang & Siau, (2019, p. 66) emphasize that people are still optimistic that artificial intelligence will open up the economy and create more jobs. However, Wang et al., believes that despite AI creating a positive impact on our lives, there are also numerous factors to consider. These factors include social norms and culture. Wang believes that with machine learning and artificial intelligence, the fourth industrial revolution has begun and is predicted to have quite a huge impact on future employment (Wang & Siau, 2019, p. 71).

3.0 Methodology

3.1 Introduction

This chapter entails various stages such as research design, which constitutes the blueprint for collection measurement and analysis of data, theoretical framework, model specification, identification of variables and how each variable is measured data type, and source.

3.2 Research Design

This research adopted a quantitative research method per the scientific paradigm. The quantitative experimental research method is suitable for identifying the real effect of technological innovations on workers who lost jobs in the manufacturing sector. The research will also be descriptive inferring national UK data to determine the growth of the manufacturing sector as a percentage of the whole economy. In addition, inferences will be drawn from secondary sources to determine the real percentage of manufacturing jobs in the UK per year.

3.3 Empirical Model

The specified empirical model for the above data collected is;

Y=Bo+B1x1+B2x2+B3x3+E

Where; Bo -Jobs lost as a result of other causes (Covid 19, retrenchment, misconduct, etc)

X1 – Jobs lost as a result of systems automation

X2- Jobs lost as a result of artificial intelligence such as robots.

X3 – Jobs lost as a result of machine learning

E – error

Y- Total number of people who lost jobs from 1990-2020.

From this mathematical model, data will be collected based on such variables and regression analysis will be performed thereafter to predict the impact of each independent variable on the endogenous variable.

3.4 Sampling Criteria

A convenience non-random sampling technique will be used to sample the population. Participants will be selected based on two criteria;

1. Worked in the manufacturing sector

2. Lost jobs in the last 20 years (1990-2020)

The sample involved 500 observations all selected from UK citizens who lost jobs in the last 20 years. The sample data sheet included; Year, Number of people who lost jobs, Number of people who lost jobs due to systems automation, Number of people who lost jobs due to machine learning, Number of people who lost due to robots, Number of people who lost as a result of other causes, Percentage of manufacturing output, and percentage of manufacturing jobs in the UK economy.

3.5 Data collection methods

After having sampled and reached a threshold of 500 participants, the data will then be collected using interviews and questionnaires. The interviews will be conducted face to face and also through the telephone only contacting willing participants. In addition, we seek to give questionnaires on various survey platforms which will be filled and emailed to us for analysis. The main focus of these interviews and questionnaires will be to identify the main cause of individuals losing jobs. The questions will also be distributed to ensure that where technology was the major cause of job loss, individuals can identify the type of technological innovation that led to their loss of jobs.

The questionnaires and interviews will involve the following parts; Year (1990-2020), Cause of job loss – Automation, Machine learning, artificial intelligence (robots), and others. The participants will only be allowed to tick only where relevant in the case of questionnaires if they meet the threshold.

The data collected will be quite crucial in assessing whether there is a significant relationship between technological progress and unemployment levels in the UK.

3.6 Data Analysis

The collected data will then be analyzed in excel, and SPSS software. A count of all participants will then be conducted based on the questionnaire and interview findings. The data will then be recorded based on the similarity of characteristics and grouped in terms of similar years for trend analysis. To assess the impact of different predictor variables, further analysis will be conducted to determine factors such as mean, variance, standard deviation, quartile, and interquartile ranges, and t hypothesis testing. Data will also be analyzed using descriptive statistics. The OLS technique was used to determine the relationship between variables (change in employment levels and predictor variables) since it’s the most efficient estimator. STATA 14.2 will be the preferred analytical tool to rank the regression model

3.7 Justification of variables

Y=Bo+B1x1+B2x2+B3x3

X1 –systems automation

This is the first independent variable. According to Goldberg (2012) automation refers to a set of activities that have been programmed. Systems automation has been one of the artificial technological advancements in recent times. Numerous researchers have linked systems automation as one of the key factors leading to employment displacement. This makes it one of the key artificial intelligence technologies to consider in the determination of the effect of technological advancements in the manufacturing sector on employment.

X2- robots.

Robots are machines that have been programmed and that can be remotely controlled or equipped with artificial intelligence enabling them to perform basic predictable functions performed by human beings. Robots’ performance greatly resembles human functions – however, robots have the upper hand since they could perform more tasks with greater precision and accuracy and still be cheaper in the long run. It is due to this reason that robots are a crucial factor when considering the effects of artificial intelligence technology on employment.

X3 –machine learning

Machine learning has been a widely used phenomenon in the determination of different trends in different data. Despite its wide usage, researchers have termed it as one of the artificial intelligence technologies that have put human employment at risk. This makes it a significant factor to consider when evaluating the technological advancements affecting employment.

3.8 Measurement of Variables

Year	Manufacturing output as %ge of the whole economy	Manufacturing jobs in the UK	Total jobs lost (Jl)	Jl due to systems automation	Jl due to robots	Jl due to machine learning	Jl due to other causes
1990	17.3	16.3	0	0	0	0	0
1991	16.9	15.7	0	0	0	0	0
1992	16.6	15.1	0	0	0	0	0
1993	16.2	14.3	0	0	0	0	0
1994	16.8	13.7	2	1	0	0	1
1995	17.1	13.8	1	0	1	0	0
1996	16.9	13.1	0	0	0	0	0
1997	16.8	13.1	1	1	0	0	0
1998	16.2	12.7	2	1	1	0	0
1999	15.6	12.4	1	0	0	0	1
2000	15	12.1	3	1	1	0	1
2001	14	11.9	6	3	2	0	1
2002	13.7	11.7	8	3	3	0	2
2003	13	11	8	1	2	1	4
2004	12.3	10.4	10	5	4	0	1
2005	11.8	9.8	9	4	3	0	2
2006	11.4	9.4	12	6	4	1	1
2007	10.8	9.2	12	4	3	2	3
2008	10.6	8.8	16	7	7	1	1
2009	10.1	8.3	21	10	9	0	2
2010	10.6	8.1	42	23	13	2	4
2011	10.6	8.1	31	21	9	1	0
2012	10.5	8.1	16	10	5	0	1
2013	10.7	7.9	12	7	3	0	2
2014	10.5	7.9	34	17	10	3	4
2015	10.4	7.7	55	26	21	2	6
2016	10.2	7.6	30	14	11	2	3
2017	10.1	7.7	8	3	2	1	2
2018	9.9	7.7	19	6	4	4	5
2019	9.8	7.6	56	24	9	7	16
2020	10.3	7.7	85	37	14	6	28

3.9 Multicollinearity

This is one of the violations of the OLS assumptions, where the explanatory variables are correlated or could be perfectly linearly correlated. The correlation coefficient (r) = 1 for Multicollinearity between explanatory variables. Multicollinearity occurs when the explanatory variable shares a common trend in time series analysis. Multicollinearity is undesirable because it produces large variances of the estimator, t ratio, they tend to be insignificant thereby giving wrong conclusions on the hypothesis tested, thus it is vital to detect the presence of Multicollinearity before carrying on the regression analysis. Formal detection can be used to establish the presence of Multicollinearity by the use of VIFs.

VIF=1/(1-R²)

Where R² is the coefficient of determination,

A VIF of 10 and above signifies the presence of Multicollinearity.

3.10 Autocorrelation

It refers to a situation where the current error term is related to the preceding error term(s). it is a common problem in time series, it occurs when the variance of the error term is sequentially interdependent. It is caused by the exclusion of relevant explanatory variables from the OLS regression model and data manipulation. In presence of autocorrelation, OLS estimators are unbiased but not efficient.

Autocorrelation can be detected by the Durbin Watson test. It is the most superior test for testing autocorrelation as compared to the Breusch Godfrey test (Sandholm & Saraniti, 2018). A DW equal to 4 is an indication of a high negative correlation level. A DW of 2 and 2.5 and values in between implies no correlation and a DW of zero indicates positive autocorrelation.

3.11 Heteroscedasticity

It is a violation of the OLS assumption where the variance of the error term changes with the magnitude of the explanatory variables (Sandholm & Saraniti, 2018). It is a common problem in cross-section data. It can be caused by several measurement errors such as the omission of important variables, the inclusion of irrelevant variables, the presence of outliers, and the manipulation of data. The Classical Linear Regression Model which assumes that the variance of the error term remains constant is violated. This will be tested by plotting the P-P plots generated from the STATA 14.2.

3.12 Stationarity Test

All variables in a Classical Linear Regression Model should have a constant mean, finite variance, and covariance, that is stationary. They should be independent of time. The unrealistic t value of the estimated coefficient is caused by non-stationary data leading to incorrect values. This calls for the stationarity test.

Chapter 4: Empirical results

4.1 Descriptive statistics

This section involves the interpretation of the empirical results of the analysis. The table below shows the summary of observations, mean, standard error, minimum and maximum values of each variable in the data set.

The study was based on a total of 31 observations from 5 variables; automation, machine learning, robots, and others. Range refers to the measure of the difference between the maximum and minimum values. For example, the minimum value for automation is 0 while the maximum is 37 hence the range is 37 (37-0).

Standard deviation is a measure of the spread of a distribution which indicates how the values of a distribution deviate from the mean. Standard deviation is vital for comparison. From the descriptive statistics table, automation has the highest standard deviation of 9.58 compared to other variables.

4.2 Correlation matrix

Correlation is the measure of the relationship between two variables (Jahrestagung, 1996, p 4). Based on the above data, there is a weak negative correlation between the predictor variables and the independent variable (unemployment rate). This implies that the predictor values are not significant factors in determining the unemployment rates in the UK market.

4.3 Diagnostic tests

4.3.1 Heteroscedasticity

This is a violation of the OLS assumption where the variance of the error term changes with the magnitude of the error term. It is a common problem in cross-section data involving non-probability sampling techniques. Heteroscedasticity is caused by the omission of important variables, wrong data manipulation, and the presence of outliers (Jahrestagung, 1996). The presence of heteroscedasticity may lead to an incorrect formula for the variance hence the t ratio may be insignificant thus resulting in wrong conclusions. However, the estimates remain unbiased.

The p-value of 0.2324 is greater than 0.05. We, therefore, reject the null hypothesis, thus indicating the absence of heteroscedasticity.

4.3.2 Autocorrelation

This refers to a situation where the current error term is related to the preceding error term. It is a common problem in time series analysis. Autocorrelation is brought about by the exclusion of relevant explanatory variables from a regression model and data manipulation especially in a time series set involving a smoothing process where the values of the error term are interrelated therefore depicting autocorrelation patterns (Jahrestagung, 1996). Even though the estimates remain unbiased, autocorrelation results in the wrong formula for variance, the t calculated is wrong and hence a wrong conclusion is made.

Since the p-value is less than 0.05, we fail to reject the null hypothesis implying that serial correlation is absent.

4.4 Multicollinearity

This occurs where some or all of the explanatory variables are related. It mostly occurs in a time series data set where the variables share a common trend or where there is a wrong model specification. Formal detection could be used to test for Multicollinearity. The most preferred is Variable Inflating Factors which is used to detect the presence of Multicollinearity.

VIF value is less than 10 thus signifying the absence of multicollinearity.

4.5 Regression

From the figure above there are 31 observations. P> F = 0.0194 is the P-value of the model and tests whether R² is different from 0 since it is lower than 0.05 thus implying that there is a statistically significant relationship between the variables. R² shows the amount of variance of poverty reduction explained by the independent variables. In this case, all the independent variables explain 35.31% of the variance in poverty reduction while 65.69% is explained by other factors outside the model. Root MSE = 1.6274 is the standard deviation of the regression since it is closer to zero. The closer to zero, the better the fit.

Two-tailed P values test the hypothesis that each coefficient is different from zero. To reject this, the P-value has to be lower than 0.05. In the above case, all the predictor values have a p-value greater than 0.05. In this case, other causes of job loss are less statistically significant as compared to automation, machine learning, and robots. The t value tests the hypothesis that the coefficient is different from zero. To test this, we need a t value greater than 1.96 (at 0.05 confidence level).

Chapter 5: Conclusion and Policy Implication

5.1 Conclusion

Based on the research, there is a low relationship between unemployment rates in the UK and technological advancements in the manufacturing sector. The regression between unemployment rates and the predictor variables is 34% and thus signifying a low positive relationship. Other factors have a huge impact on the employment rates accounting for 66% of the total factors. Based on the correlation test, it is therefore clear that technology is not a significant factor in the unemployment rates.

Based on the t statistics, it is evident that systems automation, machine learning, and robots aren’t significant factors towards unemployment rates and trends in the UK. Based on the sample selected, it is evident that other factors other than technology contributed to a larger percentage of job loss as compared to the three artificial intelligence technologies.

Based on the data analysis, it is evident that systems automation, machine learning, and robots affect the unemployment rates though in a small percentage. The technological factors are not significant in determining the rate of unemployment in the UK.

Having obtained the above results, we, therefore, join the techno-optimists and dispute the techno pessimists’ concerns as exaggerated. Based o the findings, it is quite evident that technology hasn’t been a significant cause for unemployment in the manufacturing sector. Technologies’ main aim is to improve the effectiveness at which various tasks are performed. In so doing technology increases the levels of production and leads to an expansion of these sectors thus creating new roles where substituted employees could undertake training and assume. With the rise of technological innovations and recent artificial intelligence developments, we believe that such technologies will not end up substituting human labor but rather complement human labor as the levels of production increase. The rise in technology is also likely to lead to the creation of new roles and jobs thereby reducing the unemployment levels.

5.2 Policy recommendations

Based on the results obtained from the research study, it is quite evident that technology isn’t a significant factor in determining the levels of unemployment in the UK. This study proposes the following recommendations in reducing the small impact of technology on unemployment;

1. Mandating employment

Adoption of nationwide regulations that will limit the extent to which manufacturing firms can substitute human labor for machines. The substitution effect is likely to affect the middle- and low-income earners and the government should thereby enforce measures that will mitigate the polarization effect and ensure that the future of human labor is secured.

2. Educational and training reforms

Training of employees in more nonroutine fields that are more difficult to automate. Most routine activities in the manufacturing sector can be automated and thus substituting human labor. It is therefore crucial that manufacturing firms train employees in other non-routine activities that are hard to automate.

3. Work sharing

This principle seeks to reduce the total human labor working hours in an attempt to ensure that most individuals have access to jobs.

4. Tax policies and financial incentives.

In this case, the government could offer incentives and tax cuts that will encourage the managers to favor human labor over machines (Stevens & Marchant, 2017, p. 130).

5. Small business incentives

In recent times, small and medium enterprises have been on an upward trajectory. The government should thereby offer incentives for SMEs and also enable laid-off workers with the necessary skills to start their businesses.

References

Acemoglu, D., & Restrepo, P. (2020). Robots and jobs: Evidence from US labor markets. Journal of Political Economy, 128(6), 2188-2244. https://doi.org/10.1086/705716

Bhattacharya, C. B., Sen, S., & Korschun, D. (2008). The next challenge for corporate responsibility. EFMD Insights into Sustainability and Social Responsibility, 49(2), 97-100. https://doi.org/10.5848/efmd.978-1-909201-16-3_26

Boselli, R., Cesarini, M., Mercorio, F., & Mezzanzanica, M. (2017). Using machine learning for labor market intelligence. Machine Learning and Knowledge Discovery in Databases, 330-342. https://doi.org/10.1007/978-3-319-71273-4_27

Coupe, T. (2019). Automation, job characteristics, and job insecurity. International Journal of Manpower, 40(7), 1288-1304. https://doi.org/10.1108/ijm-12-2018-0418

El Naqa, I., & Murphy, M. J. (2015). What is machine learning? Machine Learning in Radiation Oncology, 3-11. https://doi.org/10.1007/978-3-319-18305-3_1

Feng, A., & Graetz, G. (2020). Training requirements, automation, and job polarisation. The Economic Journal, 130(631), 2249-2271. https://doi.org/10.1093/ej/ueaa044

Ford, M. (2013). Could artificial intelligence create an unemployment crisis? Communications of the ACM, 56(7), 37-39. https://doi.org/10.1145/2483852.2483865

Goldberg, K. (2012). What is automation? IEEE Transactions on Automation Science and Engineering, 9(1), 1-2. https://doi.org/10.1109/tase.2011.2178910

Goos, M., & Manning, A. (2007). Lousy and lovely jobs: The rising polarization of work in Britain. Review of Economics and Statistics, 89(1), 118-133. https://doi.org/10.1162/rest.89.1.118

Goos, M., Manning, A., & Salomons, A. (2014). Explaining job polarization: Routine-biased technological change and Offshoring. American Economic Review, 104(8), 2509-2526. https://doi.org/10.1257/aer.104.8.2509

Jahrestagung, G. F. (1996). Data analysis and information systems: Statistical and conceptual approaches proceedings of the 19th annual conference of the Gesellschaft fur Klassifikation E.V. University of Basel, March 8–10, 1995. Springer.

Jordan, M. I., & Mitchell, T. M. (2015). Machine learning: Trends, perspectives, and prospects. Science, 349(6245), 255-260. https://doi.org/10.1126/science.aaa8415

King, B. A., Hammond, T., & Harrington, J. (2017). Disruptive technology: Economic consequences of artificial intelligence and the robotics revolution. Journal of Strategic Innovation and Sustainability, 12(2). https://doi.org/10.33423/jsis.v12i2.801

Margolin, V. (1989). Design Discourse: History, theory, criticism. University of Chicago Press.

Mishra, S. (2009). Educational technology: A definition with commentary – By Alan Januszewski & Michael Molenda. British Journal of Educational Technology, 40(1), 187-187. https://doi.org/10.1111/j.1467-8535.2008.00925_4.x

Nawaz Sharif, M., & Kabir, C. (1976). A generalized model for forecasting technological substitution. Technological Forecasting and Social Change, 8(4), 353-364. https://doi.org/10.1016/0040-1625(76)90027-5

Neubert, M. J., & Montañez, G. D. (2020). Virtue as a framework for the design and use of artificial intelligence. Business Horizons, 63(2), 195-204. https://doi.org/10.1016/j.bushor.2019.11.001

Parschau, C., & Hauge, J. (2020). Is automation stealing manufacturing jobs? Evidence from South Africa’s apparel industry. Geoforum, 115, 120-131. https://doi.org/10.1016/j.geoforum.2020.07.002

Peugny, C. (2019). The decline in middle-skilled employment in 12 European countries: New evidence for job polarisation. Research & Politics, 6(1), 205316801882313. https://doi.org/10.1177/2053168018823131

Piovesan, C., & Kozman, J. (2011). Cross-industry innovations in artificial intelligence. All Days. https://doi.org/10.2118/143873-ms

Robertson, D. H. (1931). Mr. Keynes’ theory of money. The Economic Journal, 41(163), 395. https://doi.org/10.2307/2223898

Rollinson, D., Buchan, A., & Choset, H. (2012). Virtual chassis for snake robots: Definition and applications. Advanced Robotics, 26(17), 2043-2064. https://doi.org/10.1080/01691864.2012.728695

Salehi, H., & Burgueño, R. (2018). Emerging artificial intelligence methods in structural engineering. Engineering Structures, 171, 170-189. https://doi.org/10.1016/j.engstruct.2018.05.084

Sandholm, W. H., & Saraniti, B. A. (2018). Vital statistics: Probability and statistics for economics and business. Oxford University Press, USA.

Sevakula, R. K., Au‐Yeung, W. M., Singh, J. P., Heist, E. K., Isselbacher, E. M., & Armoundas, A. A. (2020). State‐of‐the‐Art machine learning techniques aiming to improve patient outcomes on the cardiovascular system. Journal of the American Heart Association, 9(4). https://doi.org/10.1161/jaha.119.013924

Simmons, A., & Chappell, S. (1988). Artificial intelligence-definition and practice. IEEE Journal of Oceanic Engineering, 13(2), 14-42. https://doi.org/10.1109/48.551

Stevens, Y. A., & Marchant, G. E. (2017). Policy solutions to technological unemployment. Surviving the Machine Age, 117-130. https://doi.org/10.1007/978-3-319-51165-8_8

Waldman-Brown, A. (2020). Redeployment or robocalypse? Workers and automation in Ohio manufacturing SMEs. Cambridge Journal of Regions, Economy, and Society, 13(1), 99-115. https://doi.org/10.1093/cjres/rsz027

Wang, W., & Siau, K. (2019). Artificial intelligence, machine learning, automation, robotics, future of work, and future of humanity. Journal of Database Management, 30(1), 61-79. https://doi.org/10.4018/jdm.2019010104

Williams, A. E. (2020). Defining functional models of artificial intelligence solutions to create a library that artificial general intelligence can use to increase general problem-solving ability. https://doi.org/10.31730/osf.io/hpzb7

Zhang, Y., Sreedharan, S., Kulkarni, A., Chakraborti, T., Zhuo, H. H., & Kambhampati, S. (2017). Plan explicability and predictability for robot task planning. 2017 IEEE International Conference on Robotics and Automation (ICRA). https://doi.org/10.1109/icra.2017.7989155

Zhao, W. (2018). How to improve corporate social responsibility in the era of artificial intelligence? IOP Conference Series: Earth and Environmental Science, 186, 012036. https://doi.org/10.1088/1755-1315/186/6/012036

Smit, S., Tacke, T., Lund, S., Manyika, J., & Thiel, L. (2020). The future of work in Europe. McKinsey Global Institute.

Acemoglu, D., & Restrepo, P. (2019). 8. Artificial Intelligence, Automation, and Work (pp. 197-236). University of Chicago Press.

Green, R. A. (2014). Can “Make in India” Make Jobs? The Challenges of Manufacturing Growth and High-Quality Job Creation in India.

Appendix

Time plan

Milestone	Time	Description
Background reading	6/24/2021	This stage involves conducting in-depth research to identify a problem which the research seeks to solve.
Coming up with a topic	6/25/2021	This process involves identifying a suitable topic upon which the research intends to solve or build upon additional research.
Literature review	6/29/2021	This process involves conducting in-depth research on secondary sources and reviewing the literature of other scholars who have published scholarly sources in the area of interest.
Draft Literature submission	6/30/2021-7/1/2021	This is the initial submission that involves presenting a draft literature submission.
Data collection preparation and sampling	7/1/2021-7/3/2021	This process involves gathering the essentials required to collect data. This process also involves the distribution of questionnaires to the various software to collect data.
Actual data collection	7/4/2021	This process involves interviews both face-to-face and interviews. The process also involves collecting the questionnaires either physically or by mail.
Data analysis	7/6/2021	This process involves the utilization of various statistical software to draw inferences and also assess trends through graphical tools.
Ethics approval	7/10/2021
Coming with conclusions from the analysis	7/14/2021	After analysis, it is crucial to come up with conclusions that answer the research questions and assert whether the hypothesis is true or not.
Draft report submission	7/18/2021
Final formatting	7/20/2021	This is the final stage that entails ensuring the document is in line with laid-down guidelines.
Evaluation and review	7/20/2021	This process plays a crucial role in ensuring the report is per laid down guidelines.
Final submission	7/21/2021