How the relationship between unemployment and economic growth rate changed as the  workplace becomes increasingly automated?

Abstract

The present study aims to investigate the relationship between workplace automation and  unemployment. It is thought that an increase in productivity is indicative of increasing workplace  automation; if this were found to correspond with an increase in unemployment it could be a sign  that automation is replacing human workers. A comparative analysis of the manufacturing,

transport and construction industries in the UK revealed that growth had risen while employment  fell across all three industries. This indicates that in these sectors automation may have replaced  human workers, and that there is a lack of sufficient training and education to have had a  preventative effect. An interesting finding in the transport, communication and storage sector is  also discussed which may provide evidence for an effect of automation of growth independent of  employment.

Introduction

Increasingly automation has altered the way that unemployment rates and economic  output interact. As such any findings in this area would have far reaching consequences in current  economic theory and practice. United Kingdom statistics from the manufacturing, transport and  construction industries (long term unemployment and gross GDP) will be compared, with rates of  growth acting as a measure of increasing workplace automation, to try and draw a relationship  between these factors. The hypothesis is that a raise in economic output that does not  simultaneously reduce unemployment rates as is usually expected, could indicate that automation  of the workplace is changing the way in which unemployment and output interact, such that

traditional economic models may fail to fully explain and predict the phenomenon. It may be that  as a result of automation, unemployment is maintained at a baseline, as jobs increasingly demand  higher skill sets. There is some ambiguity in the current research, Aghion and Howitt (1990) and  

Casey (2018) seem to reflect a more up to date view of economic models of economic growth  and unemployment with regards to labour saving workplace automation. As such the current  study will attempt to elaborate on the findings of these studies and further clarify the relationship  between these factors.

Literature review

The global economy is becoming increasingly automated; we self-scan our items at the  supermarket and shop online, call centers rely on answering machines and manufacturing is  more and more reliant on automated production. While this may seem to indicate that there is  no place for human labour in many industries, it is also argued that increasingly automated  workspaces are a source of new investment and employment opportunities for highly skilled and valuable human labour. Melanie, Terry & Ulrich (2016).

The idea of an automated economy is not necessarily new, with the invention of the steam  engine came the industrial revolution. Work horses that once formed the backbone of many  industries were retired as new technology such as the combustion engine were brought in.  There was and still is widespread fear that robots will replace human workers, forcing  humans to exist on a subsistence wage, Some economists believe that the indeed during the  early 1800’s a group known as the Luddites destroyed machinery, fearing that it would come  to replace them and make them redundant. There are however some jobs that may always  require the human touch. Otekhile and Zeleny (2016)

With regard to the current state of the global market, the assumptions of Keynesian and  classical economic theories may not apply. Following the most recent global recession  unemployment has not reduced significantly, however economic growth appears to have  continued (Brynjolfsson & McAfee, 2012). Therefore, there is an anomaly that is not  explained by the two most prominent schools of economic thought. It is suggested that the  increase in growth seen without the assumed dropping in unemployment rate is the product of  new technology creating market growth but displacing workers from their jobs and therefore  failing to reduce unemployment. The interaction between economic growth and technology is  documented and its effect on unemployment is subject to much discussion.

Frey and Osborne (2017) conducted analysis across many different sectors of employment  and found that in the US 47% of jobs were at risk of being lost to automation. They also  found evidence that wages and level of education attainment strongly predicted the lack of a  computerization risk. Therefore, it could be assumed that automation of the workplace will  affect lower skilled and lower paid jobs, especially in areas like manufacture, quality  checking and information transfer and storage. Positions such as sales assistant and clerical  work were also high on the list for automation. On the other end of the scale positions such as  medical practice, therapy, hospitality and teaching were low risk. This again suggests that  automation will take over lower skilled roles and create space for investment in human  capital, effectively up-skilling workers in different or higher-level positions and sectors.  Otekhile and Zeleny (2016)

Aghion and Howitt (1990) Propose that technology exerts two factors on the rate of  unemployment with regards given to growth rate. They based their model on findings from

Pissarides’ (1999) employment growth theory that while found a link between higher  productivity and declining unemployment rates. This is a contentious link however, Layard,  Nickell and Jackman (1991) argue that unemployment rates do not consider rate of growth as  an explanatory variable, equally Phelps (1968) states that natural equilibrium employment  rates are independent of productivity growth. Regardless, Aghion and Howitt (1990) used  Pissarides’ (1990) model but adapted it as the original model failed to consider increases in  productivity can be within new job positions that replace old jobs. They found that with this  specifically applied to technological advancements that replace human workers with  automation, growth exhibited two contrasting processes on unemployment, the first they  termed capitalization, directly increased growth as a result of workplace automation, it raised  the returns of new job creation and reduced rates of unemployment. The second process,  creative discussion, describes how increased growth reduces the duration of job positions in  the labour market, this in turn has an increasing effect on unemployment rates as it indirectly  discourages the creation of new job vacancies and increases the rate at which jobs positions  are terminated. These two processes interact within the economy to create equilibrium of  unemployment that in turn influences and is influenced by growth. Thus, it suggests that the  introduction of new technology in the workplace creates stabilizing unemployment  equilibrium by creating jobs while some jobs are lost. Aghion and Howitt (1990), Otekhile  and Zeleny (2016) conclude that in the future the effect of new technologies on the workforce  is elusive, but that there is certainly an interaction that is worthy of studying as the world of  work becomes increasingly automated.

Otekhile and Zeleny (2016) discuss the impact of self-service technology on unemployment  rates and conclude that there may be an interaction between the lack of aggregate demand,

and the absence of a new sector that would absorb unemployment as a result of increasing  automation. It may be that in the future this sector could involve programming or  maintenance of these automations. Bessen (2016) found that following the automation of  textile weaving in the 19thcentury, more jobs were created in that sector as despite the  majority of the work being automatically completed the increased production reduced the  price which in turn increased the demand and created jobs in the industry not directly related  to physical manufacture. Furthermore Bessen (2016) found that in the US, computer use  negatively impacted manufacturing employment rates between 1984 and 2007 but led to a  small increase in employment in other industries. This further suggests that the interaction  between growth and productivity as a result of workplace automation and unemployment is a  complex issue that differs dependent on the industry and economic conditions that it is  implemented in.

Casey (2018) found that labour saving technology created a drop in wages and an increase in  unemployment, but also led to an increase in labour reliant technologies as a result of this, it  eventually led to a growth pattern that was similar to classical economic predictions; however  it included a level equilibrium of unemployment within this model. This seems to support the  findings of Aghion and Howitt (1990) and Pisserides (1990) that technology exerts different  processes on growth of unemployment and leads to an equilibrium level of unemployment as  a result of labor, saving work technologies. Otekhile and Zeleny (2016)

The growth trend in the EU over the last ten years indicates that following the great recession  there has been a period of stable growth, in 2014 this growth appeared to stabilize around the  2% region, however the unemployment rate within the EU up until 2014 had been growing

rapidly, peaking at around 12%, coinciding with a growth in the GDP, this suggests that the  countercyclical nature of the relationship between economic growth and unemployment may  have been affected by an exogenous factor. With regard to current research surrounding the  increasing automation of the workplace, it is hypothesized that the increase in automation has  increased growth while simultaneously increasing the unemployment rate across the EU from  2008 to 2018 (Tradingeconomics.com, 2018). It is also notable that unemployment has now  fallen to a ten year low of 8.4%, while the GDP annual growth rate is relatively stabilized at  1.90. This may suggest that new technology-based sectors may have absorbed some of the  unemployed, reducing the rate as proposed by Otekhile and Zeleny (2016). According to the  BSA software alliance, the technology and programming sector contributed 7.4% total added  value to the EU’s GDP growth over 2014, equivalent to €910 billion (Software: A €910  Billion Catalyst for the EU Economy, 2018). This would seem to support the theory that a  growing technology sector may absorb unemployment that results from the installation of  automated systems.

Method

A comparative analysis will be undertaken. Data concerning GDP growth and long-term  unemployment in the United Kingdom sectors: manufacturing, transport, and construction, will  be compared. The trends will reveal the current nature of growth and unemployment in these  industries. Data will be analyzed visually between the graphs and conclusions are drawn.

Data

The data set for the project is taken from the OECD database. The first data set shows the  growth of the gross domestic product in OECD and non-OECD countries in the period of 2008- 2018. The second data set represents the dynamics of employment change in OECD and non OECD countries in 1995-2017.

GDP per capita (nominal) is also set for each country in the study. The correlation  coefficient between GDP growth and employment rate will be compared to the GDP per capita  (nominal) in the data set. It is assumed that the countries with higher GDP per capita (nominal)  have a higher level of automation in the economy than countries with lower GDP per capita. High  final correlation coefficient will approve the hypothesis that automation is an important factor to  consider for the labor market. Otherwise, the hypothesis will not be approved.

Analysis

United Kingdom

Figure 1: Below shows the long-term unemployment rates in the United Kingdom for the  manufacturing, transport and construction industries.

The graph above shows a clear pattern in all three industries that shows unemployment in  all three industries has fallen, with a spike after the 2009 financial crisis. The trend shows the  percentage of long term unemployment falling steadily after this spike. Construction experienced  the greatest increase in long term unemployment during the 2009 financial crisis. Transport and  storage increased the least and remained relatively stable until 2014 when it began to fall.

Figure 2: Below is the performance index of growth in the manufacturing, transport, and storage  and construction industries can be seen.

In the graph above all three sectors are rising, there is a dip around the 2009 financial  crisis. Transport, storage, and communication have risen the most from 1995 to 2017, while  

manufacturing has dropped and is slowly rising again. Construction has shown a large decrease  during 2009; however, after this, the trend seems to rise steeply. Comparing the graphs, it can be  seen that while after the 2009 spike long term unemployment levels dropped relatively quickly,  GDP has not risen steeply after 2009, instead of having a gentler upward curve.

Figure 3: The graph below shows the number of persons employed in three sectors in the UK

In the graph above, the number of persons who are employed in the three sectors shows a  slowly increasing trend apart from the drop after 2008 due to the impact of the economic  recession. The manufacturing industry employs the highest number of people followed by the  transport, storage and communication sector while construction employs the least number of  people. The growing number of persons employed in the industries is related to the number of  lowering rates of unemployment and improving performance index. As the sectors performance  index improves, more people are employed leading to a decrease in the unemployment rate.

Italy

Figure 4: The long-term unemployment rates in Italy for the manufacturing, transport and  construction industries.

The unemployment rates in the three sectors increase significantly between 2008 and 2014 before  assuming a declining trend. The increase in unemployment is largely due to the effects of the  2008 economic recession. Improving unemployment rates after 2014 depict the recruitment of  more people in the three sectors. The unemployment rate is higher in the construction sector but  slightly above the other two sectors.

Figure 5: Below is the performance index of growth in the manufacturing, transport and storage  and construction industries  

The performance indices decline after 2008 in the aftermath of the economic recession  before assuming an increasing trend after 2012. The decline is however slow giving nearly flat  curves. The slow development of the three sectors is consistent with the slow increase in the  number of persons who are employed in the three sectors and the high unemployment rates in the  industry.

Figure 6: The graph below shows the number of persons employed in three sectors in Italy

The number of persons who are employed in the various sectors declines after 2008 until around  2012 due to the impact of the famous global economic recession. After 2012, the number of  people employed begins to slowly increase depicting improvement in the sector's performance.  The manufacturing industry employs significantly more persons as compared to the other two  sectors implying that manufacturing industry is an essential pillar in Italy’s economy.  

Germany

Figure 7: The long-term unemployment rates in Germany for the manufacturing, transport and  construction industries.

The unemployment rates in the three sectors decline significantly between 1995 and 2017.  This implies strong growth in the three sectors and the recruitment of more employees. The  unemployment rates show a slight increase between 2008 and 2010 due to the impact of the  economic recession. After 2010, unemployment rates decline sharply to less than 3%.

Figure 8: The graph below shows the number of persons employed in three sectors in Germany

The number of people employed in the three sectors decreases slightly with the decline in  the manufacturing industry is more pronounced as compared to the other sectors. The curves  depict a general increasing trend after 2010 although slow. The increase in the number of  employed shows improved performance in the three sectors though slow.

Figure 9: Below is the performance index of growth in the manufacturing, transport and storage  and construction industries

The performance of the three sectors declines after 2008 before increasing after 2010. The  increase in performance is related to the increase in the number of employed persons and a  decline in the rate of unemployment. Good performance in the three sectors increases the rate of  recruitment in the sectors leading to a decline in the unemployment rate.  

France

Figure 10: The long-term unemployment rates in France for the manufacturing, transport and  construction industries.

The unemployment rate in the three sectors declines up to 2008 after which there is a  slight increase in the unemployment rate to 2015 after which the rate of unemployment assumes a  declining trend. The manufacturing sector employs the largest number of people while  construction has the least. Reduced unemployment after 2015 is an indication of growth in the  economy and the three sectors.

Figure 11: The graph below shows the number of persons employed in three sectors in France

The number of people employed by the three sectors depicts an increasing trend after  2012. This comes after the economy resume from the economic financial of 2008. The increasing  number of an employed person indicates improving economy and people welfare. The increase is  consistent with the unemployment rates that decrease drastically after 2015.

Figure 12: Below is the performance index of growth in the manufacturing, transport and  storage and construction industries

According to the graph, the performance indices of the 3 sectors improve after the  economy heals from the decline caused by the 2008 economic recession. The consistent and slow  growth of performance indices is consistent with the steady increase in the number of persons  employed in the three sectors and the decline in the long-term level of unemployment.  

Results

Results failed to support the hypothesis that increased automation would lead to an  increase in productivity and an increase or maintenance in unemployment. The results found  show that long-term unemployment levels fell and performance index for each of the sector  between 1995 and 2017. The number of persons employed in each of the three sectors increases

gradually despite growing in automation and use of technology. During times of economic  recessions such as in the period after 2008, the number of employed persons declines mainly due  to the layoffs done to reduce costs in businesses. The analysis shows consistency in the trend of  unemployment rates, the number of employed and the performance of indices. Improving  performance indices enhance the ability of the three sectors to hire more people leading to  reduced unemployment rates.  

Conclusions

The results of the present study show that GDP growth causes a decrease in long-term  unemployment levels. This goes against the assumptions that growth as a possible result of  increasing automation did not increase unemployment. It may be that the findings align with  other studies that suggest a protective effect on unemployment as a result of automation. Perhaps  in these sectors, increasing productivity and automation has created space for new careers that  absorb unemployment as a result of manual workers losing their jobs to machine automation.

The delayed response of the unemployment levels to the growing sectors after the 2009  economic recession indicates a situation when the long-term unemployment remains high due to  slowed growth in the number of employed persons. This could support the idea that technology  has a preventative effect, allowing productivity to increase despite a lessened human workforce.  Perhaps the use of new technology implemented allowed the sectors to grow with little growth in  the number of employed persons. This finding could support the hypothesis that automation of  the workplace increases productivity at least partially independently of employment levels.

It is plausible that the growth of the industry has a more robust effect in generating jobs  across a variety of job roles that automation has in reducing a few select manual and specialized

roles. This may account for the lack of evidence for any mitigating effect. Indeed, the difficulty in  adequately separating the economy into industries and then the question of further splitting these  industries down into individual job descriptions is one of concern. As previously discussed, other  

studies have found that jobs that are rated as low skilled are more likely to come under threat of  automation, and jobs requiring high levels of empathy, creativity, and problem-solving are more  resistant to the threat of automation. Therefore, future studies might benefit from studying  industries in greater detail, perhaps splitting jobs into categories of skill. This may also reveal  more detailed data as a whole in the area.

One limitation of the present study is that data relating directly to the measurement of  automation in industry is not easily generated. The present study settled on measuring growth in  the industry as a whole and working from the assumption that in the current climate automation is  responsible for increasing growth rates. In the UK where manufacturing, transport, and  construction are already established industries, new bursts of growth must be at least partially  related to the implementation of automated systems. However, as there is not a direct measure, there is the possibility that confounding variables could also account for assumptions made from  the data. This issue raises a key concern or future research in addressing the area of automation  and its effect on employment. A standardized measure of automation would be extremely useful  in the analysis and prediction of the effect seen.

References

Aghion, P. and Howitt, P., 1990. A model of growth through creative destruction (No. w3223).  National Bureau of Economic Research. 

Aghion, P. and Howitt, P., 1990. A model of growth through creative destruction (No. w3223).  National Bureau of Economic Research. 

Bessen, J.E., 2016. How computer automation affects occupations: Technology, jobs, and skills. Brynjolfsson, E. and McAfee, A., 2012. Thriving in the automated economy. World Future  Society, pp.27-31.

Brynjolfsson, E. and McAfee, A., 2012. Race against the machine: How the digital revolution is  accelerating innovation, driving productivity, and irreversibly transforming employment  and the economy. Brynjolfsson and McAfee. 

Casey, G., 2018, February. Technology-driven unemployment. In 2018 Meeting Papers (No.  302). Society for Economic Dynamics. 

Layard, P.R.G., Layard, R., Nickell, S.J. and Jackman, R., 2005. Unemployment: macroeconomic  performance and the labour market. Oxford University Press on Demand. Melanie, A., Terry, G. and Ulrich, Z., 2016. The Risk of Automation for Jobs in OECD  Countries. 

Otekhile, C.A. and Zeleny, M., 2016. Self Service Technologies: A cause of  unemployment. International Journal of Entrepreneurial Knowledge, 4(1), pp.60-71. Phelps, E.S., 1968. Money-wage dynamics and labor-market equilibrium. Journal of political  economy, 76(4, Part 2), pp.678-711. 

Pissarides, C. A. (1990) Equilibrium Unemployment Theory (Oxford: Basil Blackwell).

Software: A €910 Billion Catalyst for the EU Economy. Retrieved from: 

https://softwareimpact.bsa.org/eu/pdf/EU_Economic_Impact_of_Software_Report_en.pdf  (2018). BSA.