287
Organizacija, Volume 53 Issue 4, November 2020Research Papers
1 
Received: 29th May 2020; revised: 30th August 2020; accepted: 23th September 2020
Impacts of the Transformation to 
Industry 4.0 in the Manufacturing 
Sector: The Case of the U.S.
Katarina ROJKO, Nuša ERMAN & Dejan JELOVAC
Faculty of Information Studies in Novo mesto, Ljubljanska 31a, 8000 Novo mesto, Slovenia,  
katarina.rojko@fis.unm.si, nusa.erman@fis.unm.si & dejan.jelovac@fis.unm.si
Background and purpose: The transformation to Industry 4.0 increases the number of robots installed within indus-
tries, which brings great shifts in industrial ecosystems. For this reason, our research goal was to analyze the key 
performance indicators to investigate the economic and social sustainability of the changes in production.
Methodology: The combination of official (World Bank, U.S. Bureau of Labor Statistics) and publicly available 
(Federal Reserve Economic Data, Industrial Federation of Robotics) data was used for statistical data processing, 
including comparison, correlation, cross-correlation and vector autoregression analysis, to present the past develop-
ments and also to predict future trends within the U.S. manufacturing sector.
Results: In contrast to robust industry robotization observed in the 2008–2018 period, the share of manufacturing 
output and employment declined. Nonetheless, the vector autoregression model forecast shows, that the U.S. manu-
facturing sector has arrived at a turning point, after which robotization can increase employment and labor productivi-
ty of workers, while also stimulating further growth of their education levels.
Conclusion: The transition to Industry 4.0 has a major impact on increasing demands for new knowledge and 
skills for increased productivity. Accordingly, forecasted growths of analyzed manufacturing indicators suggest that 
negative impacts of robotization in the recent past were only temporary, due to the entrance to the Industry 4.0 era.
Nonetheless, additional policies to support sustainable industry development are required.
Keywords: Industry transformation, Robotization, Industrial output, Labor productivity, Employment, Education level, 
Industry 4.0, Industry 5.0
DOI: 10.2478/orga-2020-0019
1 Introduction
This paper focuses on the recent transformation of in-
dustry, specifically the performance of the manufacturing 
sector as the most robotized sector, in light of the process 
of robotization. Namely, based on literature review and in-
itial data processing we noted that robotization has been 
accelerating during the transformation to Industry 4.0, 
while the increased number of robots within the industry 
has been having mostly negative impacts on the econom-
ic and social indicators. For this reason, our research goal 
was to analyze some of the key performance indicators to 
investigate economic and social sustainability of produc-
tion. Particularly, we aimed to investigate the first two of 
the concrete and measurable goals that are described as 
challenges and opportunities of Industry 4.0 production 
(Gianelle et al., 2016): economic sustainability of pro-
duction, social sustainability of production, production of 
future products, and environmental sustainability of pro-
duction. 
This paper intends to provide clear insight into the cur-
rent developments within Industry 4.0 transformation, due 
to the inconsistency within the existing literature, some 
highlighting positive effects of robotization, while others, 
negative ones. Namely, in some reports (e.g., in WEO, 
2018a) only positive consequences of Industry 4.0 are 
put to the front. On the other hand, there are also negative 
288
Organizacija, Volume 53 Issue 4, November 2020Research Papers
views detected in the literature (e.g., in Compagnucci et 
al., 2019). Based on an initial analysis of certain economic 
indicators, social indicators, and considering ethical con-
sequences, we first shared negative views. Namely, the 
recent transition to Industry 4.0 did not bring the expected 
outcomes, as revealed by, for example, the U.S. Bureau 
of Labor Statistics (2019a), while it brought great changes 
in society (claimed by e.g., Johannessen, 2018). Further-
more, it is predicted that all tasks that are highly manual 
and routine will soon be automated, while artificial intelli-
gence (AI) and robots might present an existential threat to 
the human role in industry.  
In contrast, governments stimulate and supply funds 
for investments in advanced technologies, including ro-
bots; they can also provide “safety nets” to enable social 
sustainability (in the form of e.g., National Robot Initiative 
2.0) of production. For this reason, it is also necessary to 
determine whether the stimulus and funds are (or will be) 
achieving their purpose. Our motivation was hence also to 
determine when (if) we can expect positive outcomes of 
investments in robotization. 
To analyze recent economic and social indicators for 
the manufacturing sector, we used official data on output, 
employment, labor productivity, education level change 
and the Industrial Federation of Robotics (IFR) data on the 
number of shipped industrial robots1. Based on the combi-
nation of these data, we presented new associations during 
the past period of Industry 4.0 and calculated the expected 
future developments. 
The following chapter of the paper includes a litera-
ture review, introducing some background views on the 
transition to Industry 4.0. The third chapter describes our 
research goal and sets out the research questions, while 
the fourth presents the data and the research methods we 
used. The next two chapters present data analysis and sum-
marize our key findings. The last chapter puts forward a 
critical view on certain literature from the field of research 
and discusses the possibilities for further human-friendly 
industry transformation. It also includes considerations 
about the options for the integration of human capabilities 
with technology, as they cannot be disregarded within this 
scope.
2 Literature Review
Due to current megatrends (globalization, global 
knowledge society, technology advancements and re-
source scarcity), the manufacturing sector is being trans-
formed on the basis of innovations that are available and 
offer competitive advantages. For this reason, the research 
priorities of Industry 4.0 are the following: efficient use 
of resources, customer-focused production, advanced pro-
duction processes, digital and virtual factories, flexible and 
smart production systems, mobility, collaboration and hu-
man-oriented production (Gianelle et al., 2016). 
Smart manufacturing is the central element of the In-
dustry 4.0 concept (Kagermann et al. in Frank et al., 2019). 
Compared to traditional production, smart factories, which 
arose within Industry 4.0, are more efficient, produce bet-
ter quality products, and enable time and cost savings. 
Smart factories, primarily at the level of processes, change 
how production plants operate to the extent that all de-
pendent production processes are transparent and combine 
the virtual and the physical (EFFRA, 2016). The products, 
resources and processes of the smart factory go into the 
composition of virtual-physical systems in which materi-
als efficiently move through the production process from 
the supplier to the finished product and further to the cus-
tomer. Smart factories combine their solutions for sharing 
and monitoring, based on a comprehensive integration of 
manufacturing facilities and technologies. For this reason, 
Jerman et al. (2020) identified the following key compe-
tencies needed by workers in smart factories (from the 
automotive industry): ICT and technical skills, innovation 
and creativity, openness to learning, adaptability, and var-
ious soft skills.
The recent technology focus in the manufacturing 
sector is directed toward factories of the future. Factories 
of the future are raising the level of robotization and are 
linking knowledge and creativity, enabling increases of 
added value per employee, and new market opportunities 
(World Economic Forum, 2018a). However, the integra-
tion of technological solutions requires a commitment to 
the comprehensive exchange of business and production 
information, based on which it is possible to increase the 
efficiency and optimization of all production processes: 
from identifying market needs, planning, modeling and 
manufacturing a new product, to planning production re-
sources, logistics, and stock management (Gianelle et al., 
1 
  1The term “industrial robots” is based on the definition of the International Organization for Standardization (ISO 8373): an 
“automatically controlled, reprogrammable multipurpose manipulator programmable in three or more axes”. Industrial robots 
can be classified according to mechanical structure: articulated, cylindrical, linear (including cartesian and gantry), parallel. or 
SCARA robots (IFR, 2020).
 2Frank et al. (2019) separate Industry 4.0 technologies into two main layers: the first is the front-end, which comprises the four 
main dimensions of Industry 4.0: smart manufacturing (including AI among others), smart products, smart supply chain and 
smart working, each of them representing a specific subset of technologies; the second layer is base technologies, which are 
those that provide connectivity and intelligence to the front-end technologies (IoT, Cloud, Big Data and Analytics).
289
Organizacija, Volume 53 Issue 4, November 2020Research Papers
2016).
Many companies have already applied Industry 4.0 
technologies2 and processes, but several authors warn 
about current threats of this fourth industrial revolution 
in the form of economic and social crises, which are, ac-
cording to Johannessen (2018), ingredients for a social 
storm. There are also other problems put forward, such as 
decreased collected labor taxes, as there is no tax and also 
no social contributions (Bottone, 2018), and other costs 
related to robots replacing human workers. There are also 
mixed views regarding, for example, a robot tax would 
hamper “freeing up” labor and sustain tension on the labor 
market in economies with labor scarcity (Vermeulen et al., 
2020), but it is advised in economies with labor surplus.
Robots replacing humans in essential sectors reduce 
or eliminate employees’ social benefits and pensions, cre-
ating social disruption (Lakshmia and Bahlib, 2020). Job 
security is threatened by a growing number of robots that 
take over even the most underpaid jobs (Johannessen, 
2018). All the routine-based and repetitive tasks, both in 
manufacturing and service-oriented professions, will be 
threatened (Lima, 2017). Moreover, robotization and in-
formatization may even contribute to a dissolution of the 
middle class (Davidow and Malone, 2014 and Chiacchio 
et al., 2018), while we should also consider the ageing of 
the workers lowering their productivity in some work-
places (e.g., machine operators and assemblers) (Aiyar et 
al. in Bogataj et al., 2019). Therefore, this development 
needs to be regulated on national and transnational lev-
els. Otherwise, employment in the manufacturing sector 
can be reduced dramatically, such as in the 1970s when 
many of the textile workers in the OECD area were made 
redundant in less than one year (Hienstra-Kupeus and Van 
Voss, 2010). Moreover, as Abramova and Grishchenko 
(2020) state: “Policy-makers should take into account the 
transition of labor from old to new jobs, reduce the period 
of adaptation, and likewise prepare education and training 
system for acquiring skills for the utilization of new tech-
nologies.”
The rapid advancements in manufacturing technology 
and in information and communication technologies re-
quire an intense and continuous update of workers’ knowl-
edge, which is essential for their integration and smooth 
adaptation into industrial working practices (Lindeberg, 
2018). Governments should require and encourage enter-
prises to organize skills-training courses and assist them to 
adapt to new technologies (Guoping et al., 2017), namely 
a rapid development of technology enables changes in in-
dustry as a whole, and the industry’s focus moved to the 
funding of advanced technological solutions to increase 
outputs and productivity. 
Although Acemoglu and Restrepo (2017) demon-
strated high and robust negative effects of robotization on 
employment and wages, and Compagnucci et al. (2019) 
used IFR data to demonstrate that the introduction of ro-
bots plays a key role in slowing down human labor and 
compensation growth, while Cho and Kim (2018) used 
IFR data for the multiple regression considering the tri-
angular relationship of employment-working-hours-wag-
es, to show that job destruction due to robotization is not 
yet very remarkable. In contrast, Cséfalvay (2020) claims 
“Recent studies clearly show that robotization is associat-
ed with economic growth and productivity gains.”
Josefsson in Lindeberg (2018) stated: “Robots are now 
everywhere, except in the productivity statistics”, hav-
ing in mind that robots do not have a positive impact on 
productivity, which can be regarded as the contemporary 
productivity paradox. Also, Lakshmia and Bahlib (2020), 
stating: “While robots and AI researchers advocate their 
significant role in productivity boost, job creation, wage 
increment and accelerated performance, such promises are 
not evident in the immediate future,” explain the observed 
discrepancies through the modern productivity paradox. 
While Glaser and Molla (2017) argue that more robots 
mean fewer jobs, the IFR (2017) has a different opinion, 
arguing: “Robots substitute labor activities but do not re-
place jobs.” and “Robots increase productivity and com-
petitiveness.” Also, the Future of Jobs 2018 survey (UK 
Parliament, 2018) indicates that new technologies create 
more jobs and new industries. However, it is not clear 
whether new technologies create more or fewer jobs than 
these technologies have eliminated. 
Some industrial robots can be very expensive, due to 
advantages including accuracy and productivity, but robots 
are still generally considered to be a cheaper workforce, 
which additionally provokes the worry and uncertainty of 
workers. Such concerns are also derived from the fact that 
changes in society can happen much slower than changes 
in technology and that people cannot compete with robots 
in the speed of data storage, capacity and retrieval of data 
(Rojko and Jelovac, 2018). People also cannot compete 
with technological solutions and robots in many other 
things: power, precision, ability to work under difficult 
conditions, and similar factors. 
As Šimek and Šperka (2019) quote in their case study: 
“The development of IT capacity and overall technology 
will move forward to a state where robots will not only 
execute the workflow but also assign the work across the 
company to both robots and humans.” The central eth-
ical concern of the transformation to Industry 4.0 is the 
decreasing need for human engagement and the lack of 
commitment to consider moral aspects. In fact, installed 
industry robots replace human workers, which leads to the 
devaluation of human work and its meaning. 
Human workers are afraid that AI and robots can be-
come an existential threat to their existence in industrial 
processes. However, there are advocates of use and de-
velopment of AI; Wilson (2017) argues that in the longer 
perspective the technical evolution, including robots, will 
serve us all, and Hendler (2017) claims “Humans working 
290
Organizacija, Volume 53 Issue 4, November 2020Research Papers
together with smart machines will be able to do better than 
either one of them can do alone.” 
In response to ethical considerations, nine top issues 
in AI were outlined (Bossemann, 2016): unemployment, 
inequality, humanity, artificial stupidity, racist robots, se-
curity, evil genius, singularity, and robot rights. For our 
research topic, unemployment, inequality, and humanity 
are ranked as the top issues, while due to the risk of un-
controlled development of AI, even the World Economic 
Forum (2018b) facing an automated future, asked: “What 
moral framework should guide us?” 
Supporters of the human role in industry believe that 
the further transformational goal should be to return jobs 
to people, but they will have to co-operate, communicate, 
and interact with advanced technological systems, includ-
ing robots. However, this poses questions about the ability 
of AI to apply ethical, cultural and moral norms of modern 
societies. Specifically, in addition to the above-mentioned 
concerns over robotization and AI, there are also other im-
pacts; for example, study results of Guo et al. (2019) sug-
gest that humanoid robot’s emotional behaviors can evoke 
significant emotional responses among users. Accordingly, 
as already widely acknowledged, Putilo et al. (2020) claim 
that the mass robotization of the manufacturing and ser-
vice sectors requires solving the problems of the necessity 
of introducing the status of electronic personality to intelli-
gent robots, defining their legal capacity in the civil, labor, 
and other fields of law.
The different opinions presented above show the incon-
sistency within the literature and provide a blurry picture 
of the current transformation of industry. Moreover, since 
governments stimulate and provide funds for investments 
in advanced technologies, it is necessary to determine 
whether those stimulus and funds are (or will be) achiev-
ing their purpose and if economic and social sustainability 
(as two of the most important goals of Industry 4.0 accord-
ing to Gianelle et al. (2016)) of the production are viable 
outcomes of the recent industry robotization or not.
3 Research goal and research 
questions
Addressing the challenges of transformations within 
Industry 4.0 should be a major part of the research in this 
field, as the effects of robotization are currently not clear. 
For this reason, our research goal was to analyze the key 
performance indicators to investigate the economic and 
social sustainability of production. We used the combina-
tion of the official and publicly available data to present 
developments within the past researched period and to cal-
culate further trends. 
We, therefore, set the following research questions:
1)How do employment and output change in the man-
ufacturing sector in relation to the industry as a whole?
2)What were the impacts of robotization on employ-
ment and labor productivity in manufacturing during the 
period of entrance to Industry 4.0? 
2.1)How is the impact of robotization reflected in em-
ployment and labor productivity in manufacturing? 
3)How did the employees’ education level change dur-
ing the period of manufacturing entrance to Industry 4.0?
3.1)Is the impact of robotization reflected in the change 
of education level of employees in manufacturing?
4)What are the expected future interrelated develop-
ments of manufacturing labor productivity, employment, 
education level and robotization?
In the next section, we present the methodology with 
which we made comparisons and calculations to answer 
the above-listed research questions.
4 Methods
We used official and publicly available data to answer 
our research questions, as such data allow the highest relia-
bility and lead to solid conclusions. An analysis and statis-
tical data processing of the U.S. Bureau of Labor Statistics 
(2019a, 2019b and 2019c), the Federal Reserve Economic 
(2018), and World Bank (2019) data, combined with pub-
licly available IFR (2018b and 2019) data and sources, 
enabled us to develop an alternative view on the current 
transformation of the industry ecosystem. 
We focused on the data for the period between 2008 
and 2018. We started by presenting the data on employ-
ment and output in U.S. industry as a whole, compared to 
the same data for the U.S. manufacturing sector, as well as 
the predicted growth thereof in the forthcoming 10-year 
period. In the next stage of our analysis, we focused on 
multiple time-series data containing the number of indus-
trial robot shipments, labor productivity in the manufactur-
ing sector, employment and employees’ education level in 
manufacturing. Here, the labor productivity is measured as 
the amount of goods and services produced according to 
the number of hours worked to produce those goods and 
services (U.S. Bureau of Labor Statistics, 2019b), both in 
the manufacturing sector. 
We first outlined the number of industrial robot ship-
ments and employment in manufacturing and compared 
them to labor productivity in the manufacturing sector. 
Afterwards, we also studied general education trends vs. 
industrial robot shipment. In both cases, the significances 
of trends as a function of time were tested using regression 
analysis and t-tests.  
Following the demonstrated trends, we further inves-
tigated the association between the observed time-series 
data. On the one hand, we calculated the correlation be-
tween labor productivity in the manufacturing sector and 
the number of industrial robot shipments, as well as be-
291
Organizacija, Volume 53 Issue 4, November 2020Research Papers
tween labor productivity and the number of employees 
in manufacturing. On the other hand, we also analyzed 
the association between education level change and the 
number of industrial robot shipments. To do so, we used 
cross-correlation analysis, which, besides quantifying the 
immediate correlation between the two time series, ena-
bles assessing the responsiveness of one time series to the 
other, by simultaneously taking into account the time di-
mension (t). The cross-correlations are calculated as a nor-
malized version of the cross-covariance function (Shum-
way and Stoffer, 2017):
(1)
where xt and yt are the observed time series with N 
data points, μx and μy are their means, σxx(0) and σyy(0) 
are their variances, T represents a time lag, and σxy(T) is 
the cross-covariance function between the observed time 
series at lag T.
Following the normalized cross-covariance function, 
we calculated the cross-correlation function (CCF), which 
represents a set of correlations between the time series 
xt+T and the time series yt for lag values T=0,±1,±2, and 
so on. The value of  r_xy ranges from  ̶ 1 to 1. It is to note 
also that the cross-correlation function is not symmetric 
around zero, which means that r_xy (T)≠r_xy (-T). When 
lag is negative (T<0), the set of correlations refers to the 
correlation between the x time series at a time before t (x_
(t-T)), and the y time series at time t (y_t). When the lag is 
positive (T>0), the set of correlations refers to the corre-
lation between the x time series at a time after t (x_(t+T)), 
and the y time series at time t (y_t). When time lag equals 
zero (T=0), the value of cross-correlation equals the value 
of the Pearson correlation coefficient:
(2)
where σxy is the covariance between the observed 
time series xt and yt, while σx and σy are their standard 
deviations. Significant correlation values suggest that the 
changes in one time series appear immediately following 
the changes in the other time series (Shin, 2017).
To present the results of cross-correlation analysis, we 
used a graphical representation of CCF using correlation 
plots. The x-axis represents a time lag (T), and the y-axis 
represents the value of cross-correlation. Blue dotted lines 
(at Figures 2, 3, and 5) represent the approximate 95% 
confidence intervals that serve as the threshold for iden-
tifying statistically significant cross-correlation values. To 
identify a potential significant correlation between lagged 
time series x and time series y at time t, we examined the 
values of CCF for negative lag values (T < 0). In this case, 
when the cross-correlation is statistically significant, then 
the change in time series x leads to the change in time se-
ries y after the period, determined by the lag T, at which 
cross-correlation is significant (Shin, 2017).
Based on the cross-correlation analysis results between 
the observed time-series, we also examined possible dy-
namics of the observed time series. Thus, we finally con-
ducted vector autoregression (VAR) analysis, which is 
an extension of well-known linear- and auto-regressions. 
When analyzing k time series, the VAR analysis allows 
predicting all k time series variables using a single mod-
el, instead of fitting k regression models individually. The 
VAR model is best presented as a system of multiple equa-
tions, in which there is one equation for each time series 
as a dependent variable, where the goal is to estimate each 
equation separately using the lagged values of all k time 
series as predictors (Hanck et al. 2019). Following the ba-
sic form of a VAR model (as suggested by Zivot and Wang, 
2006), for the i-th time series in the VAR model, we can 
present the equation as:
(3)
where k represents the number of time series included 
in the model, T denotes a time lag, and the βs are estimated 
using ordinary least squares.
To fit the best model, we also had to estimate the op-
timal number of lags (T), which defines a lag order of the 
VAR model. Here we used the Akaike’s Information Crite-
ria (AIC) and chose the specification that has the smallest 
value for AIC (as explained by Zivot and Wang, 2006). We 
also assessed the quality of the defined VAR model using 
F-test and its p-value, as well as the multiple R2, to iden-
tify the proportion of the variance explained by the model. 
Then, we estimated the proposed VAR model, in which 
the significance of the results was tested using a t-test. Fi-
nally, we used the estimated model to predict and present 
the future trends in labor productivity in the manufacturing 
sector, the number of industrial robot shipments, manufac-
turing employment, and manufacturing employees’ educa-
tion level change. 
292
Organizacija, Volume 53 Issue 4, November 2020Research Papers
5 Results
In comparison to previous studies, we decided to ana-
lyze data for one of the most robotized economies (the 
U.S.)3 and attempted to consider several of the most rele-
vant indicators simultaneously, to show a concise and clear 
image of the performance of the manufacturing sector in 
light of the transformation to Industry 4.0. Specifically, 
as Essentra (2019) is suggesting, the U.S. is the leading 
market in Industry 4.0, while Schreiber in Essentra (2019) 
explains, “The U.S. government sees manufacturing as 
an engine for growth and we’ve seen increased research 
and development (R&D) tax credits and lower corporate 
rates.”
1 
   3In terms of industrial robot shipments, China led among individual countries, based on 36.5% share in 2018 (1.4 percentage po-
ints less than in 2017), while the U.S. climbed from fourth to third place (behind Japan and Republic of Korea) in 2018, based on 
21.6% y-o-y growth (IFR, 2019). The driver behind the continued growth of industrial robot shipments in the U.S. is the ongoing 
trend to automate production in order to strengthen the country’s industry and to keep manufacturing production at home, or to 
bring it back home (IFR, 2018a). In 2018, the average global robot density of 99 industrial robots installed per 10,000 employees 
was measured in the manufacturing industry. With an average of 114 units, Europe is the region with the highest robot density, 
while the Americas the second with 99 units, followed by Asia/Australia with 91 units (IFR, 2019).
 2008 2018/2008 2018 2028/2018 2028
Total Employment 149,276.00 7.88% 161,037.70 5.22% 169,435.90
Manufacturing Employment 13,405.50 -5.35% 12,688.70 -5.05% 12,048.00
Total Output 28,909.50 14.99% 33,241.90 20.47% 40,045.30
Manufacturing Output 5,989.20 2.42% 6,134.00 16.81% 7,164.90
Table 1: Employment (in thousands) and output (billions of dollars) - companies from the manufacturing sector and industry 
as a whole in the U.S. (years 2008 to 2028) (U.S. Bureau of Labor Statistics, 2019a)
In Table 1, we present data on employment and out-
puts in companies from the manufacturing sector and in 
industry in general in the U.S. to answer our first research 
question. The data show that the employment in the U.S. 
grew at a compound annual growth rate (CAGR) of 0.7% 
from 2008 until 2018 in the industry as a whole, while in 
the manufacturing sector it declined at a CAGR of 0.5%. 
Projections show that in the next decade we can expect 
CAGR growth of 0.5% in industry as a whole, while in 
manufacturing a continued decline of 0.5% is projected. 
Furthermore, the data from Table 1 reveal that from 
2008 to 2018, the U.S. industrial output increased by 
15.0% in general, while in manufacturing it increased only 
by 2.4%. Considering the same (constant 2011 dollars) 
variable, from 2008 to 2018, the CAGR of 1.3% in the 
industry as a whole was measured, while in the manufac-
turing sector it grew only by 0.2%. From 2018 to 2028, 
stronger CAGR growth of 1.7% in industry as a whole 
is expected, while in manufacturing only the increase of 
1.4% is projected. 
As a result, the percentage distribution among differ-
ent sectors shows (U.S. Bureau data, 2019a) that the share 
of output from the manufacturing sector is decreasing, as 
it declined from 20.7% in 2008 to 18.5% in 2018, and a 
further decline is projected in 2028 to 17.9%. Similarly, 
the share of manufacturing employees declined from 9.0% 
to 7.9% (2008 vs. 2018), while further decline to 7.1% is 
expected in 2028. 
Data from Figure 1 answer our second research ques-
tion, as they present the number of industrial robot ship-
ments in the U.S., which grew at a CAGR of 11.3% from 
2008 to 2018 (IFR, 2018b and 2019). As shown in Table 
2, the growth in the number of industrial robot shipments 
in the 2008–2018 period is significant, for which time ex-
plains 93.4% of the variation in growth.
In contrast, in the same period, employment in man-
ufacturing declined at a CAGR of 0.5%, and as shown 
in Table 2 the change in the number of employees in the 
2008–2018 period is not significant. Additionally, time ex-
plains only 2.5% of the variation in the number of employ-
ees. Similarly, labor productivity in manufacturing grew at 
a CAGR of 0.8% from 2008 to 2018, while the growth in 
labor productivity in this period is not significant and time 
explains only 6.7% of the variation in labor productivity 
growth.
We can observe the drop in both the number of indus-
trial robot shipments as well as the employment in man-
ufacturing in 2009 as compared to 2008, which can be 
attributed to the 2008-economic crisis. In the following 
years, from 2011 to 2018, we ascertain steady increases 
in both the number of industrial robot shipments and em-
ployment in manufacturing, but the increase in the latter 
was much less intense.
293
Organizacija, Volume 53 Issue 4, November 2020Research Papers
Figure 1: Number of shipped industrial robots and number of employees vs. labor productivity (percentage change from pre-
vious year) in the U.S. manufacturing sector (2008-2018) (IFR, 2018b and 2019, U.S. Bureau of Labor Statistics, 2019b and 
2019c)
RS Estimate Std. error
t-value F
R2statistic p-value statistic p-value
Constant 6017.3818** 1744.9017 3.449 0.007
126.308 0.000 0.934
Time 2891.3909** 257.2717 11.239 0.000
EMPL
Constant 12073.7000** 348.6529 34.630 0.000
0.234 0.640 0.025
Time 24.8796 51.4061 0.484 0.640
PROD
Constant 1.7145 1.3463 1.273 0.235
0.650 0.441 0.067
Time -0.1600 0.1985 -0.806 0.441
Table 2: Number of shipped industrial robots, number of employees and labor productivity (percentage change from previous 
year) in the U.S. manufacturing sector (2008-2018) as functions of time
** The estimate is significant at the 0.01 level.
Furthermore, labor productivity in manufacturing var-
ied from one year to another (U.S. Bureau of Labor Sta-
tistics, 2019b), but we would like to emphasize that the 
situation was not improving. If we exclude direct impacts 
of the economic crisis starting in 2008, the labor produc-
tivity in manufacturing was very low, especially in the last 
eight years presented, in which the greatest year-over-year 
growth of 1.4% was recorded in 2013, and the strongest 
decline of 1.6% was measured in 2015.
To answer our first (2.1) research sub-question, we 
used the cross-correlation analysis to test the association 
and potential impact of industrial robot shipments on em-
ployment and labor productivity in the manufacturing sec-
tor in the observed 2008–2018 period (Figure 2 and Figure 
3). 
Figure 2 presents the CCF for the number of industrial 
robot shipments and employment in manufacturing. The 
correlation between the observed time-series data (when 
time lag = 0 years) shows weak positive association (r = 
0.24), which means that immediately after the number of 
shipped robots increases, the employment in manufactur-
ing increases too. However, significant correlation appears 
294
Organizacija, Volume 53 Issue 4, November 2020Research Papers
only at time lag =  ̶ 1 year (r = 0.60), indicating that an 
increase in the number of shipped robots more certainly 
leads to an increase in manufacturing employment about 
one year later. We can also support this finding with the 
calculation that 97.1% of the variation (R2 = 0.971) in 
manufacturing employment is explained when predicting 
it using the one-year lagged number of shipped robots. 
Figure 2: Cross-correlation function (CCF) for the number of industrial robot shipments and employment in manufacturing in 
the U.S. in the observed 2008-2018 period 
Figure 3 presents the CCF for the number of indus-
trial robot shipments and labor productivity in the man-
ufacturing sector in the observed 2008–2018 period. The 
correlation between the observed time-series data (when 
time lag = 0 years) shows moderate negative association (r 
= -0.34), which means that immediately after the number 
of shipped robots increases, the labor productivity in the 
manufacturing sector decreases. Moreover, at time lag =  ̶ 
1 year the negative correlation becomes even stronger (r = 
-0.54) indicating that an increase in the number of shipped 
robots about one year later leads to an even stronger de-
crease in labor productivity. Only after three years (when 
time lag = -3 years) the correlation turns out to be positive 
but very weak (r = 0.15). However, it is noteworthy that in 
the presented CCF none of the correlations is statistically 
significant, which indicates that we have to recognize that 
the reasons behind weak labor productivity growth can be 
several and different and not only related to the introduc-
tion of robots. We can also support this finding with the 
calculation that only 11.6% of the variation (R2 = 0.116) 
in labor productivity is explained when predicting it using 
the number of shipped robots.
Data from Figure 4 answers our third research ques-
tion, presenting the number of industrial robot shipments 
in the U.S., and the employees’ education level4 in the 
manufacturing sector expressed in proportions (IFR, 
2018b and 2019, WB, 2019). The proportion of manu-
facturing employees with lower levels of education (i.e., 
below upper secondary and upper secondary) is steadily 
decreasing, while the proportion of manufacturing em-
ployees with the highest level of education (i.e., tertiary) 
is constantly increasing. In 2015, the proportion of man-
1 
 4To present a level of education the source (WB, 2019) uses the term “adults'” for employees aged 24-64, while in our paper we 
use simply “employees’ education level”.
295
Organizacija, Volume 53 Issue 4, November 2020Research Papers
Figure 3: Cross-correlation function (CCF) for the number of industrial robot shipments and labor productivity in manufac-
turing sector in the U.S. in the observed 2008-2018 period
Figure 4: Number of shipped industrial robots vs. level of manufacturing employees’ education (% of 24-64-year-olds) in the 
U.S. in the observed 2008-2018 period (IFR, 2018b and 2019, WB, 2019)
296
Organizacija, Volume 53 Issue 4, November 2020Research Papers
Below Upper secondary Estimate Std. Error
t-value F
R2statistic p-value statistic p-value
Constant 0.1169** 0.0014 85.079 0.000
106.572 0.000 0.922
Time -0.0021** 0.0002 -10.323 0.000
Upper secondary
Constant 0.4833** 0.0012 395.706 0.000
557.300 0.000 0.984
Time -0.0043** 0.0002 -23.607 0.000
Tertiary
Constant 0.3999** 0.0019 211.100 0.000
515.646 0.000 0.983
Time 0.0063** 0.0003 22.708 0.000
Table 3: Number of shipped industrial robots vs. level of manufacturing employees’ education (% of 24-64-year-olds) in the 
U.S. in the observed 2008-2018 period as functions of time
** The estimate is significant at the 0.01 level.
ufacturing employees with the upper secondary and ter-
tiary levels of education is nearly the same, but since 2016 
the proportion of those with the tertiary level of education 
exceeds the proportion of those with the upper secondary 
level of education.
As shown in Table 3, for all three levels of education, 
the trends as functions of time are also significant, where 
time explains 92.2%, 98.4%, and 98.3% of the variation in 
the below upper secondary, upper secondary and tertiary 
levels of education, respectively.
We are aware that the level of education depends on 
many factors, and it is growing evenly in most countries, 
regardless of the introduction of robots. Nonetheless, we 
decided to analyze the interrelation among the levels of 
education in manufacturing and robotization, since ro-
botization requires more educated employees (Guoping 
et al., 2017; Lima, 2017; Lindeberg, 2018; Abramova and 
Grishchenko, 2020), which are able to supervise, install, 
reconfigure, etc. robots. We also wanted to determine if 
robots mostly replace jobs of the employees with the mid 
education levels (as stated by Davidow and Malone, 2014 
and Chiacchio et al., 2018) or with the lowest one. In this 
regard, Figure 4 data indicate that when the number of in-
dustrial robot shipments increases, the education level of 
manufacturing employees also increases, while the strong-
er decline of manufacturing employees is measured among 
those with upper secondary education, not among those 
with below upper secondary education.
Since there is an obvious trend in the change5 of man-
ufacturing employees’ level of education as compared to 
the number of robot shipments, we also tested the associa-
tion between the two indicators using the cross-correlation 
analysis (Figure 5) to answer our second (3.1) sub research 
question, although we emphasize that we keep in mind that 
the mutual association between the two observed variables 
is not decisive. 
The correlation between the observed time-series 
data (when time lag = 0 years) shows significant and very 
strong positive correlation (r = 0.96), which means that 
immediately after the number of shipped robots increases, 
the level of manufacturing employees’ education also in-
creases. The data also show significant correlation at time 
lag =  ̶  1 year (r = 0.75) indicating that one year after 
the number of shipped robots increases, the manufactur-
ing employees’ education level still increases. In contrast, 
a significant correlation also appears at time lag = 1 year (r 
= 0.69), which implies that the increase in manufacturing 
employees’ education level can be observed even one year 
before the number of shipped robots increases. 
Although we have emphasized that it is not decisive 
that growth in the level of education can be attributed to 
the introduction of robots, the fact that 92.5% of the varia-
tion (R2 = 0.925) in manufacturing employees’ education 
level change is explained by the number of industrial robot 
shipments is self-evident. 
Following the findings presented above, there is a ten-
dency of association between labor productivity, employ-
ment and industrial robot shipments on the one hand, and 
the education level and industrial robot shipments on the 
other. 
1 
 5Here, we calculated the change in education level as the difference between the proportion of manufacturing employees with 
below upper secondary level of education and the proportion of those with tertiary level of education.
297
Organizacija, Volume 53 Issue 4, November 2020Research Papers
Figure 5: Cross-correlation function (CCF) for the number of industrial robot shipments and change in manufacturing employ-
ees’ education level in the U.S. in the observed 2008-2018 period
Hence, we generated a time-series using multiple var-
iables, i.e., labor productivity in the manufacturing sector 
(PROD), industrial robot shipments (RS), employment in 
the manufacturing sector (EMPL), and change in the ed-
ucation level of manufacturing employees (EDU_ch). To 
eliminate the influence of different measurement scales of 
the observed variables, we used the standardized time-se-
ries data from 2008 to 2018 and conducted VAR analy-
sis to predict multiple time series variables, using a sin-
gle model. Here, we used the AIC criteria to find the best 
model and to estimate the optimal number of lags for the 
time-series. Based on the 1-ordered time lag is the most 
optimal; therefore, we estimated the vector autoregression 
model specifying the time lag order as 1. 
Using VAR model analysis, we were looking to esti-
mate four autoregression equations, one for each of the 
observed variables: 
The results are presented in Table 4 and Table 5; when 
combined with the trends presented in Figure 6, they an-
swer our last research question: “What are the expected fu-
ture interrelated developments of manufacturing labor pro-
ductivity, employment, education level, and robotization?”
When predicting each of the observed variables, the 
VAR models are statistically significant (p < 0.05). It also 
turns out that the strength of the analyzed relationships is 
relatively high. Namely, the values of the R2 range from 
0.865 and 0.986, which indicates that at least 86.5% of the 
variation in dependent variables is caused by independent 
variables in the VAR model.
298
Organizacija, Volume 53 Issue 4, November 2020Research Papers
Table 4: Summary of vector autoregression models
F
R2 Adjusted R2
Model statistic df p-value
1a 8.04 4, 5 0.0210 0.865 0.758
2b 51.89 4, 5 0.0003 0.977 0.958
3c 86.85 4, 5 0.0000 0.986 0.975
4d 53.21 4, 5 0.0003 0.977 0.959
a Dependent variable: PROD; Independent variables: RS, EMPL, EDU_ch.
b Dependent variable: RS; Independent variables: PROD, EMPL, EDU_ch.
c Dependent variable: EMPL; Independent variables: PROD, RS, EDU_ch.
d Dependent variable: EDU_ch; Independent variables: PROD, RS, EMPL.
PROD Estimate Std. error t-value p-value
Constant 0.1296 0.1684 0.770 0.476
PRODt-1 -0.4119 0.2144 -1.921 0.113
RS t-1 -3.7311
** 0.7378 -5.057 0.004
EMPL t-1 -0.0732 0.1995 -0.367 0.729
EDU_ch t-1 3.3002
** 0.7790 4.237 0.008
RS
Constant 0.2551* 0.0666 3.832 0.012
PRODt-1 -0.0139 0.0848 -0.164 0.876
RS t-1 0.3589 0.2917 1.230 0.273
EMPL t-1 -0.3196
** 0.0789 -4.052 0.010
EDU_ch t-1 0.7051 0.3079 2.209 0.071
EMPL
Constant -0.1039* 0.0366 -2.839 0.036
PRODt-1 0.0219 0.0466 0.470 0.658
RS t-1 0.4054 0.1603 2.528 0.053
EMPL t-1 0.1280
* 0.0434 2.952 0.032
EDU_ch t-1 0.3381 0.1693 1.998 0.102
EDU_CH
Constant 0.3120 0.0646 4.832 0.005
PRODt-1 -0.0279 0.0822 -0.339 0.748
RS t-1 -0.0814 0.2830 -0.288 0.785
EMPL t-1 -0.0725 0.0765 -0.948 0.387
EDU_ch t-1 1.1782
* 0.2987 3.944 0.011
Table 5: Vector autoregression coefficients for the estimated models 
** The estimate is significant at the 0.01 level.
* The estimate is significant at the 0.05 level. 
299
Organizacija, Volume 53 Issue 4, November 2020Research Papers
In the first model, when estimating the equation for 
labor productivity, the first time lag of industrial robot 
shipments and the first time lag of change in education 
level for manufacturing employees are statistically signif-
icant, which indicates that labor productivity significantly 
drops one year after the number of industrial robot ship-
ments increases. Furthermore, the results also show that 
the increase in the educational level leads to a significant 
increase in labor productivity one year later. It also turns 
out that the first time lag of labor productivity in the man-
ufacturing sector and the first time lag of the number of 
employees in manufacturing has some impact on labor 
productivity, although their impact is not statistically sig-
nificant. 
In the second model, when estimating the equation for 
the number of industrial robot shipments, the first time lag 
of the number of employees in manufacturing is statisti-
cally significant. This finding indicates that the increase in 
the number of employees in manufacturing has a negative 
influence on the number of industrial robot shipments one 
year later. 
In the third model, when estimating the equation for 
employment in manufacturing, the first time lag of em-
ployment in manufacturing is statistically significant. 
In the last model, when estimating the equation for the 
change in education level for manufacturing employees, 
the first time lag of the change in education level for man-
ufacturing employees is statistically significant.
Figure 6: Predicted values for U.S. labor productivity (PROD), industrial robot shipments (RS), number of employees (EMPL) 
and change in employees’ education level (EDU_ch) in manufacturing vs. industrial robot shipments (RS) based on VAR model 
(2008-2018 historical, 2019–2028 predicted)
300
Organizacija, Volume 53 Issue 4, November 2020Research Papers
Following the estimated VAR models, we have also 
conducted the predictions for the future, which are pre-
sented in Figure 6. The left side of the figure shows the ac-
tual data for the observed 2008–2018 period, and the right 
side represents the predictions for the next 10-year period 
based on each autoregression equation. The predicted val-
ues for each individual time-series variable are presented 
as lines, and the shaded areas represent 80% and 95% con-
fidence intervals thereof.
The results show that in the coming 10-year period, 
we can expect an increase in all of the observed variables. 
Accordingly, although the number of industrial robot ship-
ments will increase, we can also expect a growth in the 
number of employees in the manufacturing sector with 
higher levels of education as compared to 2018. Conse-
quently, labor productivity in the manufacturing sector 
will increase as well. These calculations project a bright 
future, which is in contrast to (for example) the U.S. Bu-
reau of Labor of Statistics (2019a) data presented in Figure 
1. Specifically, Figure 1 data project employment decline 
of 5.05% in manufacturing over the forecasted period 
(2028/18). Nonetheless, in case of significant econom-
ic situation changes in the forecasted decade, the future 
might not be as bright as predicted in Figure 6.
6 Discussion
It is widely accepted that the current transition to In-
dustry 4.0 requires significant and also long-term invest-
ments in company resources, while new technological 
opportunities change the performance of industrial en-
terprises and offer possibilities for increased profits and 
labor productivity. Nonetheless, besides investments in 
technology, investments in workers are also required. New 
employee profiles have to upgrade their skills all the time 
and have to be able to co-operate and work together with 
robots that take over traditional jobs within industry. 
In this paper, we have attempted to provide clearer 
insight into the current developments within Industry 4.0 
transformation, while our motivation was also to determine 
when (if) can we expect better impacts of investments in 
robotization, based on analysis of recent economic and so-
cial indicators for the manufacturing sector.
Answers and reflections are given to our research ques-
tions below:
1) How has the employment and output changed in the 
manufacturing sector in relation to the industry as a whole?
During the transition to Industry 4.0, U.S. manufactur-
ing output has barely grown, while the number of employ-
ees decreased (Table 1). Having these findings in mind, 
the observed significant investments in the multipurpose 
robots in the U.S. (Figure 1) did not prevent manufacturing 
output share decline within industries, while they also lead 
to the reduction of manufacturing employees’ share within 
industries. 
Simpler tasks with lower required qualifications are 
taken by machines, robots. However, we have to keep in 
mind that robots can be extremely helpful in workplaces 
with difficult conditions and where the possibility of health 
problems occurs. Nonetheless, since people cannot com-
pete with robots in many things (power, etc.), we agree 
with Dyer et al. (2010) that creativity will be one of the 
most sought-after talents in the future. Besides creativity, 
reasoning, adaptability, emotional intelligence, moral act-
ing, critical thinking, problem solving, decision making, 
and similar, are the human qualities that will continue to 
gain importance. Thus, both now and in the future, there 
will be a constantly growing need for new knowledge and 
competences, and thus the place for human workers.
2) What were the impacts of robotization on employ-
ment and labor productivity in manufacturing during the 
period of entrance to Industry 4.0?  
As the number of industrial robot shipments in the U.S. 
grew from 2008 to 2018, employment in manufacturing 
declined in 2018 vs. 2008 (Figure 1). Nonetheless, em-
ployment in manufacturing declined year-on-year only in 
2009 and 2010, which can be attributed to the 2008-eco-
nomic crisis, while in the period from 2011 to 2018 it was 
steadily increasing, although much less intense, as the in-
dustrial robot shipments.
Figure 1 data also present unexpectedly low labor pro-
ductivity growth in the U.S. manufacturing sector, com-
pared to robust growths of robots’ shipments, especially in 
the last eight years, and as such they support the claims of 
Josefsson in Lindeberg (2018), that robots are now every-
where, except in productivity statistics.
For this reason, we must agree with Guoping et al. 
(2017) that government has to play a catalytic and super-
visory role in processes of industry transformation and to 
accelerate and simultaneously guarantee proper develop-
ment. Namely, the agitations over Industry 4.0, in which 
people who are replaced by robots have to be considered, 
and the preparation of the new national and transnation-
al policies for sustainable industry development must be 
stimulated. Otherwise, as robots are becoming less expen-
sive and more capable, they might replace many human 
workers, while unemployment could grow to meet the 
conditions for social crisis (even a social storm, as claimed 
by Johannessen, 2018).
2.1) How is the impact of robotization reflected in em-
ployment and labor productivity in manufacturing? 
To answer this research question, we used the 
cross-correlation analysis to test the association and poten-
tial impact of industrial robot shipments on employment 
and labor productivity in the U.S. manufacturing sector in 
the observed 2008–2018 period (Figure 2 and Figure 3). 
The correlation between the observed time-series data 
(when time lag = 0 years) presented in Figure 2 shows a 
weak, positive association, which means that immediately 
after the number of shipped robots increases, the employ-
ment in manufacturing increases too. However, significant 
301
Organizacija, Volume 53 Issue 4, November 2020Research Papers
correlation appears only at time lag =  ̶ 1 year, indicating 
that an increase in the number of shipped robots more cer-
tainly leads to an increase in manufacturing employment 
about one year later. We also were able to support this 
finding with the calculation that 97.1% of the variation in 
manufacturing employment is explained when predicting 
it using the one-year lagged number of shipped robots. 
In contrast, the correlation between the observed 
time-series data (when time lag = 0 years) presented in 
Figure 3 shows a moderate, negative association, which 
indicates that immediately after the number of shipped 
robots increases, the labor productivity in manufacturing 
sector decreases. Moreover, at time lag =  ̶ 1 year the neg-
ative correlation becomes even stronger, implying that an 
increase in the number of shipped robots about one year 
later leads to an even stronger decline in labor productivi-
ty. Only after three years did the correlation turn positive, 
but very weak. However, it is to be noted that in this cal-
culation none of the correlations is statistically significant, 
which indicates that we have to recognize that the reasons 
behind weak labor productivity growth can be several and 
different and not only related to the introduction of robots. 
We can also support this finding with the calculation that 
only 11.6% of the variation in labor productivity is ex-
plained when predicting it using the number of shipped 
robots. Thus, our research results cannot firmly support 
either the advocates of contemporary productivity para-
dox (Josefsson in Lindberg, 2018; Lakshmia and Bahlib, 
2020), the positive claims of IFR (2017) “robots increase 
productivity” nor Cséfalvay’s claim (2020) “Today in the 
literature a new consensus is emerging that adoption of 
industrial robots considerably increases productivity and 
contributes significantly to economic growth.”
3) How did the employees’ education level change dur-
ing the period of manufacturing entrance to Industry 4.0? 
We determined that the proportion of manufacturing 
employees with lower levels of education (i.e., below up-
per secondary and upper secondary) is steadily decreasing, 
while the proportion of manufacturing employees with the 
highest level of education (i.e., tertiary) is constantly in-
creasing (Figure 4) and follows the robot shipment growth. 
We also determined that in the past researched period in 
the U.S., robots more likely replaced employees with the 
mid education levels (as stated by Davidow and Malone, 
2014 and Chiacchio et al., 2018), since the stronger decline 
in the number of manufacturing employees was measured 
among those with upper secondary education, not among 
those with below upper secondary education.
Despite the above-indicated findings, we are aware that 
the level of education in manufacturing depends on many 
factors, not only on the introduction of robots. Nonethe-
less, robotization requires more educated employees (as 
suggested by Guoping et al., 2017; Lima, 2017; Lindeberg, 
2018; Abramova and Grishchenko, 2020).
3.1) Is the impact of industry robotization reflected in 
the change of education level of employees in manufac-
turing?
The correlation between the observed time-series data 
(when time lag = 0 years), presented in Figure 5, reveals 
significant and very strong positive correlation, which 
means that immediately after the number of shipped robots 
increases, the level of manufacturing employees’ educa-
tion also increases. The data also show significant corre-
lation at time lag =  ̶  1 year, indicating that one year after 
the number of shipped robots increases, the manufacturing 
employees’ education level still increases. In contrast, sig-
nificant correlation appears at time lag = 1 year as well, 
implying that the increase in manufacturing employees’ 
education level can be observed even one year before the 
number of shipped robots increases. Following these find-
ings, we assume that the introduction of robots to the work 
process is at least three-fold, since a year before, in the 
same year and in the year after the investments in new ro-
bots in manufacturing, the employees’ education increases 
at a significant level.
We nonetheless emphasize that the introduction of ro-
bots is not decisive for the level of education change, but 
the calculations show that 92.5% of the variation in man-
ufacturing employees’ education level change is explained 
by the number of industrial robots’ shipment. From this 
point of view, the results support the claims that roboti-
zation tends to require higher educated workers (Guoping 
et al., 2017; Lima, 2017; Lindeberg, 2018; Abramova and 
Grishchenko, 2020).   
4) What are the expected future interrelated develop-
ments of manufacturing labor productivity, employment, 
education level and robotization?
Based on the estimated VAR models (Figure 6), we 
predict positive developments in the U.S. manufacturing 
sector. Bearing in mind the presented calculations, it might 
be anticipated, that the U.S. has now finally arrived at the 
turning point within the Industry 4.0 period, after a period 
of negative trends of employment and labor productivity, 
which was the impact of country’s entrance into Industry 
4.0. 
The forecasted growth in manufacturing employment, 
labor productivity, education, coexistent with the robot 
shipment growth, suggests that outlined negative trends 
of the number of employees and labor productivity in the 
researched period were only temporary, and we believe 
that was due to the lack of opportunity and knowledge on 
how to exploit robotization for the maximum efficiency. 
These findings oppose the U.S. Bureau of Labor of Sta-
tistics (2019a) data presented in Table 1 and also disagree 
with the claims of other authors (e.g., Compagnucci et al., 
2019). 
We emphasize that the projected positive future devel-
opments are expected for the U.S., while the same could 
also hold for most other developed countries in terms of 
Industry 4.0 adoption, as the calculations were made based 
on data for the U.S., which are one of the most robotized 
economies, according to IFR (2019). Nonetheless, in case 
302
Organizacija, Volume 53 Issue 4, November 2020Research Papers
of significant economic changes in the forecasted period 
(e.g., impact of current coronavirus pandemic (COV-
ID-19), the future might not be as bright as projected in 
Figure 6.
Especially when analyzing and predicting such com-
plex and impactful trends of economic and social stabili-
ty, it is necessary to consider different factors and models, 
although projections are usually made based on a single 
observed variable. For this reason, we used VAR analysis, 
which allows predicting several time series variables using 
a single model. Although this analytical technique enables 
a more advanced forecasting approach than, for example, 
a simple linear regression method, it still assumes that the 
trend growths in the analyzed time series are all linear, 
which is also reflected in the predictions we made for the 
coming 10-year period. However, we recognize that unex-
pected events, such as the current coronavirus pandemic, 
also cause changes and cyclical trends in the studied in-
dicators. As shown on the left side of Figure 6, the past 
trends of the economic indicators used are more or less 
non-linear, which is the most obvious in case of labor pro-
ductivity, whereas in the case of the number of employees 
the trend is relatively stable. In this regard, it is proposed 
for future research that the linear predictive approach using 
VAR modeling is done on the transformed time series data 
(we can use, e.g., logarithmic or power transformations), 
since the transformations may inhibit greater fluctuations 
and improve linearity in non-linear time series (Shumway 
& Stoffer, 2017). Future research could furthermore use 
some other approaches, which allow analyzing non-linear 
multiple time series, e.g., non-linear VAR models (as de-
scribed by Kilian & Lütkepohl, 2017) or even VAR neural 
network models (as described by Yasin et al., 2018), which 
allow the analysis of non-linear time series without their 
prior transformation. 
There are also other limitations of our research, which 
are suggested to be addressed in future research, such as 
analyzing a longer period, which could lead to more accu-
rate statistical calculations and forecasts. In contrast, since 
our focus was on Industry 4.0, it would not make any sense 
to analyze a longer research period.
In future research, however, we recommend a compar-
ison of the U.S. market with other countries (considering 
different demographic structure, etc.), and other details 
about the robot shipments are advised to be analyzed (e.g., 
their value, type, application etc.), as this would provide 
additional insight into the impacts of transformation to In-
dustry 4.0. In light of the current coronavirus pandemic, 
which significantly changed the criteria and approach to 
robotization, it would be especially interesting to analyze 
how manufacturing companies already equipped with In-
dustry 4.0 technology and robots will overcome the sit-
uation in comparison with other “lagging” companies. 
Since robots can work in surroundings harmful for human 
health, they should have a major advantage; nonetheless, 
generally decreased demand for manufactured goods most 
likely also has a negative impact on them.
7 Conclusion
During the transition to Industry 4.0, the manufactur-
ing output, as well as the number of employees in manu-
facturing and labor productivity, have barely grown in the 
U.S. (see Table 1 and Figure 1). Nonetheless, our projec-
tions for the next decade (Figure 6) project brighter de-
velopments; for this reason, we support advocates of the 
development and use of AI and robots (as Wilson, 2017 
and Hendler, 2017) in manufacturing. The future is in co-
operation between robots and human. This partnership can 
bring prosperity and increased labor productivity, which is 
expected within a future global economy. 
However, it is self-evident that people must maintain 
their decisive role in work processes; otherwise, robots 
might replace human workers, which would lead to even 
greater social inequity and consequent crisis. One can ask: 
“How can Sophia get human rights (Saudi Arabia citizen-
ship), but I can’t keep my job to maintain basic human 
dignity?” For this reason, new strategic directions are 
required, in which the humanization of industry must be 
highlighted as the most important. This implies that the 
transformation of industry has to be planned sustainably, 
due to the possible significant negative impact on society 
as a whole. For this reason, there is a constant need to in-
vestigate the impacts of industrial development, since the 
changes in society are much slower than the changes in 
technology. 
The data we used for our research showed that there is 
still room for a human contribution in industry. Specifical-
ly, based on the VAR model, in the coming 10-year period, 
we can expect an increase in all of the observed variables. 
Projections thus show that while the number of industrial 
robot shipments will increase, we can also expect a growth 
in the number of employees in the manufacturing sector 
with increasing level of education, and the growth of la-
bor productivity in the U.S. manufacturing sector. These 
calculation results are contrary to certain data and findings 
we present in our paper (e.g. U.S. Bureau of Labor of Sta-
tistics (2019a), Acemoulgu and Restrepo (2017), Glaser 
and Molla (2017), Josefsson in Lindeberg (2018), Laksh-
mia and Bahlib (2020), while they support findings from 
the IFR (2017), the UK Parliament (2018), and Cséfalvay 
(2020), among others. However, as already mentioned, 
in case of any disruption or changed circumstances, the 
calculated indicators for the future might change. None-
theless, since the transition to Industry 4.0 has a major 
impact on increasing demands for new knowledge about 
technologies and production processes, at least a part of 
our modelled predictions should hold. 
Especially now, at the “turning point” of the U.S. man-
303
Organizacija, Volume 53 Issue 4, November 2020Research Papers
ufacturing sector, as indicated by our calculations, workers 
must focus on other tasks: those enabling the optimization 
of work processes, including control, analytics and regu-
lation, while emphasis must also be on the effective com-
munication, creativity and problem-solving acting. Here, 
the opportunity for an increased role of human workers 
should appear, while a higher degree of reasoning, flexi-
bility, reliability, problem-solving acting, moral responsi-
bility, and fast learning will be expected from the employ-
ees. According to Essentra (2019), for example, there is a 
current lack of the right workforce composition and the 
skill sets needed for the future; the skills U.S. manufactur-
ers are most lacking are technology and computer skills, 
problem-solving ability, design engineering and mathe-
matics, claiming “U.S. manufacturers are currently more 
concerned about how to attract and upskill workers rather 
than making plans to let them go.” 
Among the above-mentioned challenges and opportu-
nities in the field of industry robotization, the economic 
and social sustainability of production requires the integra-
tion of human and technology. As Harper (2019) explains, 
a human-computer interaction perspective can help define 
interactions between AI and users that can enhance rather 
than substitute creativity. The new step in industry devel-
opment – Industry 5.0 – should therefore focus primar-
ily on human and robot engagement and the integration 
of human knowledge, creativity, intuition, skills, experi-
ence, etc. within robotized production. Moreover, Industry 
5.0 should give reasons for human presence in industrial 
processes, but in forms based on the changed staffing re-
quirements of modern production, including the ability to 
co-operate with robots. 
Thus Industry 5.0 is supposed to change robots from 
programmable machined into ideal human companions, 
commonly termed as “cobot”, that will already know, or 
quickly learn, what to do (Nahavandi, 2019). Although the 
prediction of Industry 5.0 is currently questionable, fur-
ther industry transition should be regulated and oriented 
towards the integration of the virtual and the physical, con-
sidering the changed role of qualified workers and the new-
ly established social responsibility of robots. Accordingly, 
we conclude that there is a need for increased awareness 
about the consequences of the transformation to Industry 
4.0, which calls for new policies for industry’s econom-
ic and social sustainability, considering the specifics and 
labor (surplus or scarcity (Vermeulen, 2020) demands of 
individual economies and industrial sectors.
Literature
Abramova, N., & Grishchenko, N. (2020). ICTs, Labour 
Productivity and Employment: Sustainability in Indus-
tries in Russia. Procedia Manufacturing, 43, 299-305. 
https://doi.org/10.1016/j.promfg.2020.02.161 
Acemoglu, D., & Restrepo, P. (2017). Robots and jobs: Ev-
idence from US labor markets. Retrieved from https://
www.nber.org/papers/w23285 
Bogataj, D., Battini, D., Calzavara, M., & Persona, A. 
2019. The Response Latency in Global Production 
and Logistics: A Trade-off between Robotization 
and Globalization of a Chain. Procedia Manufactur-
ing. 39, 1428-1437. https://doi.org/10.1016/j.prom-
fg.2020.01.309 
Bossemann, J. (2016). Top 9 ethical issues in artificial in-
telligence. Retrieved from https://www.weforum.org/
agenda/2016/10/top-10-ethical-issues-in-artificial-in-
telligence/ 
Bottone, G. 2018. A tax on robots? Retrieved from http://
www.finanze.it/export/sites/finanze/it/.content/Docu-
menti/Varie/dfwp3_2018.pdf 
Chiacchio, F., Petropoulos, G., & Pichler, D. (2018). 
The impact of industrial robots on EU employ-
ment and wages: A local labour market approach. 
Bruegel Working Paper, 2018/02, Bruegel, Brus-
sels. Retrieved from https://www.econstor.eu/bitstre
am/10419/207001/1/1028792522.pdf 
Cho, J., & Kim, J. (2018). Identifying Factors Reinforc-
ing Robotization: Interactive Forces of Employment, 
Working Hour and Wage. Sustainability. 10(2), 490. 
https://doi.org/10.3390/su10020490 
Compagnucci, F., Gentili, A., Valentini, E., & Gallegati, 
M. (2019). Robotization and labour dislocation in the 
manufacturing sectors of OECD countries: a panel 
VAR approach, Applied Economics, 51:57, 6127-6138. 
https://doi.org/10.1080/00036846.2019.1659499 
Cséfalvay, Z. (2020). Robotization in Central and Eastern 
Europe: catching up or dependence? European Plan-
ning Studies. 28(8), 1534-1553. https://doi.org/10.108
0/09654313.2019.1694647 
Davidow, W. H., & Malone M. C. (2014). What happens 
to society when robots replace workers? Harvard 
Business Review. December 10, 2014. https://hbr.
org/2014/12/what-happens-to-society-when-robots-re-
place-workers 
Dyer, J., Gregersen, H., & Christensen., C. M. (2010). The 
innovator’s DNA: mastering the five skills of disruptive 
innovators. Boston: Harvard Business Review Press.
EFFRA. (2016). Factories 4.0 and Beyond. Recommenda-
tions for the work programme 18-19-20 of the FoF PPP 
under Horizon 2020. Retrieved from http://www.effra.
eu/sites/default/files/factories40_beyond_v31_public.
pdf 
Essentra. (2019). A guide to Industry 4.0 in the US. (23. 
Jan.). Retrieved from  https://www.essentracompo-
nents.com/en-us/news/guides/a-guide-to-industry-40-
in-the-us 
Federal Reserve Economic Data (2018). Manufacturing 
Sector: Labor Productivity. Retrieved from https://
fred.stlouisfed.org/series/MPU9900063  
304
Organizacija, Volume 53 Issue 4, November 2020Research Papers
Frank, A., Dalenogare, L., & Ayala, N. (2019). Industry 
4.0 technologies: Implementation patterns in manu-
facturing companies. International Journal of Produc-
tion Economics, 210, 15-26. https://doi.org/10.1016/j.
ijpe.2019.01.004 .
Gianelle, C., Kyriakou, D., Cohen, C., & Przeor, M., eds. 
(2016). Implementing Smart Specialization: A Hand-
book. Brussels: European Commission. 
Glaser, A., & Molla, R. (2017). The number of robots sold 
in the U.S. will jump nearly 300 percent in nine years. 
Job-stealing robots are steadily taking over America. 
Recode, 3 April 2017. Retrieved from https://www.
recode.net/2017/4/3/15123006/robots-sold-ameri-
ca-growth-300-percent-jobs-automation 
Guo F., Li, M., Qu, Q., & Duffy, V. G. (2019). The Effect 
of a Humanoid Robot’s Emotional Behaviors on Users’ 
Emotional Responses: Evidence from Pupillometry 
and Electroencephalography Measures. Internation-
al Journal of Human–Computer Interaction. 35(20), 
1947-1959. https://doi.org/10.1080/10447318.2019.1
587938 
Guoping, L., Yun, H., & Aizhi, W. (2017). Fourth Indus-
trial Revolution: Technological Drivers, Impacts and 
Coping Methods. Chinese Geographical Science. 
2017, 27(4), 626–637. http://doi.org/10.1007/s11769-
017-0890-x 
Hanck, C., Arnold, M., Gerber, A., & Schmelzer, M. 
(2019). Introduction to Econometrics with R. Depart-
ment of Business Administration and Economics, Uni-
versity of Duisburg-Essen, Essen, Germany.  Retrieved 
from https://www.econometrics-with-r.org/ 
Harper, R. H. (2019). The Role of HCI in the Age of AI. In-
ternational Journal of Human–Computer Interaction. 
35(15), 1331-1344. http://doi.org/10.1080/10447318.2
019.1631527 
Hendler, J. (2017). Fear Not, AI May Be Our New Best 
Partners in Creative Solutions. Retrieved from https://
www.techemergence.com/fear-not-ai-may-be-our-
new-best-partners-in-creative-solutions-a-conversa-
tion-with-dr-james-hendler/ 
IFR (2017). The Impact of Robots on Productivity, Em-
ployment and Jobs, a positioning paper by the Inter-
national Federation of Robotics, April 2017. Retrieved 
from https://ifr.org/img/office/IFR_The_Impact_of_
Robots_on_Employment.pdf  
IFR (2018a). Executive Summary World Robotics 2017 
Industrial Robots. Retrieved from https://ifr.org/down-
loads/press/Executive_Summary_WR_2017_Industri-
al_Robots.pdf  
IFR (2018b). Executive Summary World Robotics 2018 
Industrial Robots. Retrieved from https://ifr.org/down-
loads/press2018/Executive_Summary_WR_2018_In-
dustrial_Robots.pdf  
IFR (2019). Executive Summary World Robotics 2019 
Industrial Robots. Retrieved from https://ifr.org/down-
loads/press2018/Executive%20Summary%20WR%20
2019%20Industrial%20Robots.pdf 
IFR (2020). Industrial Robots. Retrieved from https://ifr.
org/industrial-robots 
Jerman, A., Bertoncelj, A., Dominici, G., Pejić Bach, M., 
& Trnavčević, A. (2020). Conceptual key competency 
model for smart factories in production processes. Or-
ganizacija, 53(1), 68-79. https://doi.org/10.2478/orga-
2020-0005 
Johannessen, J. A. (2018). Automation, Innovation and 
Economic Crisis, Surviving the Fourth Industrial Rev-
olution. London: Routledge.
Kilian, L., & Lütkepohl, H. (2017). Structural Vector Au-
toregressive Analysis. Cambridge:  Cambridge Univer-
sity Press.
Lakshmia, V., & Bahlib, B. (2020). Understanding the ro-
botization landscape transformation: A centering reso-
nance analysis.  Journal of Innovation & Knowledge. 
5(1), 59-67. https://doi.org/10.1016/j.jik.2019.01.005 
Lima, P. U. (2017). Autonomous mobile robotics: a system 
perspective. New York: CRS Press. 
Lindeberg, R. (2018). Robots Are Now Everywhere, Ex-
cept in the Productivity Statistics. Bloomberg, 10. 
April 2018. Retrieved from https://www.bloomberg.
com/news/articles/2018-04-10/robots-are-now-every-
where-except-in-the-productivity-statistics 
Nahavandi, S. (2019). Industry 5.0—A Human-Centric 
Solution. Sustainability, 11 (16), 4371. https://doi.
org/10.3390/su11164371 
Putilo, N. V., Volkova, N. S., & Antonova, N. V. (2020). 
Robotization in the Area of Labor and Employment: 
On the Verge of the Fourth Industrial Revolution. In: 
Popkova E. & Sergi B. (eds.). Artificial Intelligence: 
Anthropogenic Nature vs. Social Origin. ISC Confer-
ence - Volgograd 2020. Advances in Intelligent Sys-
tems and Computing, 1100. Springer, Cham.
Rojko, K., & Jelovac, D. (2018). Human role in factories 
of the future. Zbornik radova DIEC 2018. 1(1):165-
178. http://diec.ipi-akademija.ba/zbornici-radova/ 
Shin, Y. (2017). Time Series Analysis in the Social Scienc-
es: The Fundamentals. University of California Press, 
Oakland, California.
Shumway, R. H., & Stoffer, D. S. (2017). Time Series 
Analysis and Its Applications: With R Examples (4th 
ed.). Springer, Cham, Switzerland.
Šimek, D., & Šperka, R. (2019). How Robot/human Or-
chestration Can Help in an HR Department: A Case 
Study From a Pilot Implementation. Organizacija, 
52(3). https://doi.org/10.2478/orga-2019-0013 
UK Parliament (2018). Government response to House of 
Lords Artificial Intelligence Select Committee’s Re-
port on AI in the UK: Ready, Willing and Able? Re-
trieved from https://www.parliament.uk/documents/
305
Organizacija, Volume 53 Issue 4, November 2020Research Papers
lords-committees/Artificial-Intelligence/AI-Govern-
ment-Response2.pdf 
U.S. Bureau of Labor Statistics. (2019a). Industry Output 
and Employment. Retrieved from https://www.bls.
gov/emp/data/industry-out-and-emp.htm 
U.S. Bureau of Labor Statistics. (2019b). Division of Ma-
jor Sector Productivity. Retrieved from https://www.
bls.gov/lpc/  
U.S. Bureau of Labor Statistics. (2019c). All Employees, 
Manufacturing [MANEMP]. Retrieved from FRED, 
Federal Reserve Bank of St. Louis. https://fred.stlou-
isfed.org/series/MANEMP 
Vermeulen, B., Pyka, A., & Saviotti, P. P. (2020). Robots, 
Structural Change, and Employment: Future Sce-
narios. Springer. https://link.springer.com/content/
pdf/10.1007/978-3-319-57365-6_9-2.pdf  
Wilson, M. (2017). Implementation of Robot System. New 
York: Butterworth-Heinemann.
World Bank (2019). World Bank open data. Economy & 
Growth, https://data.worldbank.org/ 
World Economic Forum (2018a). Europe, Asia Lead the 
Way to the Factories of the Future. Retrieved from 
https://www.weforum.org/press/2018/09/europe-asia-
lead-the-way-to-the-factories-of-the-future/ 
World Economic Forum (2018b). The Future of Jobs Re-
port. Retrieved from http://www3.weforum.org/docs/
WEF_Future_of_Jobs_2018.pdf 
Yasin, H., Warsito, B., Santoso, R., & Suparti, S. (2018). 
Soft Computation Vector Autoregressive Neural Net-
work (VAR-NN) GUI-Based. E3S Web of Conferences, 
73, p.  13008.
Zivot, E., & Wang, J. (2006). Vector Autoregressive Mod-
els for Multivariate Time Series. In: Modeling Finan-
cial Time-Series with S-Plus, 385-429. New York: 
Springer. 
Katarina Rojko is an Assistant Professor at the Faculty 
of Information Studies in Novo mesto, Slovenia. In 
research she focuses on information society and 
e-business, Industry 4.0., and also on innovative 
learning and teaching in higher education.
Nuša Erman is an Assistant Professor in the field 
of quantitative methods at the Faculty of Information 
Studies in Novo mesto, Slovenia. Her research work 
is focused on statistical methods and quantitative 
research, especially in the fields of information and 
social sciences. 
Dejan Jelovac is a Professor of organizational sciences 
and business ethics at the Faculty of Information 
Studies in Novo mesto, Slovenia. His research interests 
include values and virtues in organizational culture, 
entrepreneurship, management and leadership, 
cross-cultural comparisons of organizational culture 
in business and public administration, challenges 
in multicultural business communication, and 
organizational development and corporate social 
responsibility in small and medium-sized business.
Vplivi transformacije v industrijo 4.0 v proizvodnem sektorju: primer ZDA
Ozadje in namen: S transformacijo v industrijo 4.0 se v industriji povečuje število nameščenih robotov, kar prinaša 
velike premike v industrijskih ekosistemih. Zato je bil naš raziskovalni cilj analizirati ključne kazalnike uspešnosti, da 
bi raziskali ekonomsko in socialno vzdržnost teh sprememb v proizvodnji.
Zasnova / metodologija / pristop: Kombinacija uradnih (World Bank, U.S. Bureau of Labor Statistics) in javno 
dostopnih (Federal Reserve Economic Data, Industrial Federation of Robotics) podatkov je bila uporabljena za stati-
stično obdelavo podatkov, vključujoč primerjavo, korelacijo, navzkrižno korelacijo in analizo vektorske avtoregresije, 
da bi predstavili pretekli razvoj in tudi napovedali prihodnje trende v ameriškem proizvodnem sektorju.
Rezultati: V nasprotju z močno robotizacijo v obdobju 2008–2018 se je delež proizvodnje in zaposlitve v proizvod-
nem sektorju v tem obdobju zmanjšal glede na celotno industrijo. Kljub temu napoved modela vektorske avtoregre-
sije kaže, da je ameriški proizvodni sektor prišel do prelomne točke, po kateri lahko robotizacija poveča zaposlenost 
in produktivnost delavcev, hkrati pa spodbuja nadaljnjo rast njihove izobrazbene ravni.
Zaključek: Prehod na industrijo 4.0 močno vpliva na vse večje potrebe po novih znanjih in veščinah za večjo pro-
duktivnost. V skladu s tem napovedane rasti analiziranih proizvodnih kazalnikov kažejo, da so bili negativni vplivi 
robotizacije v nedavni preteklosti le začasni zaradi vstopa v industrijo 4.0. Kljub temu pa so potrebne dodatne politike 
za podporo trajnostnemu razvoju industrije.
Ključne besede: Transformacija industrije, Robotizacija, Industrijska proizvodnja, Produktivnost dela, Zaposlitev, 
Stopnja izobrazbe, Industrija 4.0, Industrija 5.0