Digitalization and Productivity: Evidence on Indonesian Large and Medium Industries

: This article estimates the relationship between digitalization and labor productivity in the manufacturing sector of a developing country represented by Indonesia. Literature investigating the effect of digitalization on labor productivity in the manufacturing sector is still rare in developing countries. Thus, the present study contributes to filling the gap regarding digitalization and labor productivity in developing countries. It used data of 33,577 large and medium industrial companies in Indonesia in 2017 obtained from the Large and Medium Industry Survey 2017 database, Statistics Indonesia. This study used econometric regression in cross-sectional data with labor productivity as the dependent variable and ICT asset proportion as the primary independent variable, representing the digitalization variable. The findings revealed that digitalization is positively and significantly associated with labor productivity in Indonesia's large and medium industries.


I. INTRODUCTION
Digitalization has become necessary in the business sectors to compete globally in this Industry 4.0 era.The digital transformation may disrupt the status quo business model in many sectors such as banking, retail, transportation, education, health, energy, and industry.On the positive side, literature discussed that recent technologies might generate many advantages for the industrial process, such as efficient and flexible manufacturing [1] [2], adaptation to decision making [1], and sustainable industry.[3] Yet, most employers are worried about the labor force turmoil if they tried to invest in digital technology, which is suspected as a potential replacement for some workers, i.e., technicians, operators, and administrators.Rifkin (2014) stated that the digital revolution would drastically decrease employment as the marginal costs of production are close to zero in the long-term [4].Additionally, considering financial aspects, most companies tend to invest their budget in more urgent needs than digital technologies due to budget limitations.A relatively small number of manufacturing companies exploit the benefit of Information and Communication Technology (ICT) in their business processes, particularly in developing countries, due to the uncertain return of digital technology investment in the future.With limited knowledge and financial capacity, the rate of digitalization in developing countries seems slower than that of developed ones at the macro-level.[5] Bell and Pavitt (1993) showed that developed countries' digitalization process is already at the development stage of innovation for technological aspects.[6] In most developing countries, it is still at the adoption stage, including Indonesia.To anticipate the issue, The Government of The Republic of Indonesia has launched a roadmap, namely Making Indonesia 4.0, to disseminate the benefits of digitalization throughout the nation and create a conducive atmosphere for Indonesia's manufacturing sector, especially in terms of implementing digital technology.[7] The effort will be made by ensuring abundant, cheap, and continuous electricity sources, accelerating digital infrastructure development, including high speed and wide-coverage internet access and digital capabilities, encouraging incentives on technology investments, and improving the human resource quality with government cooperation.Regarding the consequences of work-related issues, this guideline explains a scenario of fundamental structural change.The total labor demand may not necessarily decline, strongly changing in structure, e.g., machines may replace jobs or specific tasks.However, there might be new opportunities for affected workers instead.However, digitalization depends on Government intervention and location, types of industry, and own industry resources.The positive results coming from developed countries may differ in developing countries, especially there must be a high cost incurred with limited financial capacity that most industries in Indonesia are faced with.Thus, the correlation between digitalization investment and productivity in a developing country needs to be discussed.Numbers of studies discover that digitalization can leverage labor income using company productivity enhancement in developed countries.[ developing countries is minimal, not using the econometric approach in most cases.[17] [18] To the best of our knowledge, there is still a limited number of quantitative studies analyzing the effect of digitalization on labor productivity in Indonesia.There is no study using ICT proportion as the primary explanatory variable to identify digitalization's association with labor productivity in developing countries.Thus, this study aimed to estimate the relationship between digitalization and labor productivity in Indonesia's manufacturing sector.Looking at previous literature, the hypothesis in this study was that digitalization is positively associated with Indonesian labor productivity in large and medium manufacturing industries.
The paper contributes to filling the gap, especially the growth of literature in Indonesia, by providing evidence on digitalization in Indonesian manufacturing companies.It also used industry-level latest data in 2017 in manufacture.The manufacturing industry was chosen as it plays a significant role in Indonesia's Gross Domestic Product (GDP).The shock or change in manufacture may significantly improve the Indonesian economy.Additionally, this study may support the government in providing the right policies for manufacturing in Indonesia to compete with other countries and provide a scientific reference for industrial companies in terms of digital technology implementation to increase labor productivity.This paper is organized as follows.Section 1 provides an introduction, while Section 2 summarizes the literature review.Section 3 focuses on the modeling consisting of empirical model, variable description, and data source.Section 4 provides results and discusses the findings.The last section summarizes the results and concludes the research.

II. LITERATURE REVIEW
Digitalization correlates strongly with information and telecommunication technologies (ICT).UNESCO (2006) defines ICTs as forms of technology used to transmit, store, create, share, or exchange information.[19] The technological equipment and resources referred to computers and their systems, networks, live broadcast or monitoring technology, recorded technology, telephones, and satellites.Other technologies drive digitalization, particularly in the manufacturing sector, i.e., sensor technology and Big Data and Analytics.[20] In terms of productivity, its definition continues to develop over time, making it not as simple as merely describing the ratio of production efficiency.According to Asian Productivity Organization (2015) and Rasmussen (2013), productivity is defined as the ratio of produced product (output) value with production resources (input) value in the production process, calculated from the output value divided by the input value.[21] [22] In principle, digitalization will benefit companies in terms of time and cost savings, increased production flexibility and quality of the product, and human resource.[23] These benefits will ultimately leverage the company's efficiency and competitiveness so that its revenue will increase despite using the original amount of inputs.Let assume that there is no significant addition in labor input.It can be considered that an increase in revenue is interpreted as an increase in the company's labor productivity.Implementing ICT technologies enables companies to save more time, energy, and materials for production processes.[15] As a result, workers can complete their tasks faster, so the company will increase its production capacity.However, there have been several mixed results regarding the correlation between ICT implementation and productivity in empirical studies.Identifying the effect of ICT and research and development (R&D) investment from Indian manufacturing, Khanna and Sharma (2018) found a strong relationship between ICT and R&D investment in generating labor productivity growth, even across different sub-sample.[14] Meanwhile, using long-term panel data, Camina et al. (2020) found that the effect of digitalization depends on which technology was to be assessed.[3] Robotics and efficient production systems may improve productivity, yet other computerization did not boost or explain productivity.A weak signal of correlation is also found in Leviakangas et al. (2017) in the Australian construction sector.[12]

III. MODEL AND DATA A. Model
The present study's estimation model includes all variables used and their operational definitions (shown in Equation 1).They will be estimated using a cross-sectional regression model transformed in logarithmic values.where i is the cross-sectional index unit (companies).LP represents the labor productivity of the companies, which is measured by the logarithmic value of the company's revenue divided by several labors (million rupiahs), as referred to in the literature.The export dummy variable measures the distinction between the exporter and nonexporter company categories, EXP, 1 for exporting company and 0 if otherwise.[14] [15] The study also distinguished companies based on their foreign or local ownership measured by the company ownership dummy, OWN, which is 1 for full or partial foreign ownership and 0 if otherwise.[14] SZ describes the company's size, measured by a dummy variable, 1 for a large company (100 or more employees) and 0 if otherwise.[10] [15] [36] The CAP_INTSV variable describes the company categorization based on capitalor labor-intensive characteristic, which is 1 for the capital-intensive company and 0 if otherwise.[37] To control industry-particular attributes, the study also categorized the type of industrial subsectors as its control variables as in investigations by Doms et al. ( 2004) and Kroll (2018).[10] [15] It controls for all 24 subsectors in Indonesia based on Indonesia Standard Industrial Classification Codes as attached in Appendix A. [38] The study included all 23 subsectors dummy variable, 1 for the company's relevant subsector and 0 if otherwise.As the base subsector, the study chose the coal and petroleum refinery industry products in its subsector dummy variable estimation (see Appendix A).To test its robustness model, it designed estimation based on variables used in all six models, as shown in Table 3. Model 1 consisted of the least variables, while Model 6 consisted of the complete ones.

B. Data
The study used secondary data obtained from the Large and Medium Industries Survey 2017, published by Statistics Indonesia in 2019.It used the latest data set of Large and Medium Survey conducted in 2017 and 33,577 large and medium industrial companies (micro-level) in Indonesia.[39] IV.EMPIRICAL RESULTS AND DISCUSSION Table 1 and Table 2 show the descriptive statistics of the variables in the model.Table 1 displayed that the average revenue of large and medium industrial companies in Indonesia was Rp. 223 billion in 2017.Furthermore, the average number of workers in large and medium industrial companies in that year was 197 people from all sub-sectors.The minimum number of the worker was 20 people, while the maximum was 55,252 people.Then the labor productivity variable showed an average value of Rp. 1,286 billion per person from all companies.The average ICT asset value of large and medium industrial companies in Indonesia was Rp. 171 million in 2017.Some companies did not invest funds in ICT infrastructure or equal to Rp. 0 for the value of ICT assets from the survey results.Meanwhile, the highest value allocated by companies for ICT infrastructure investment was Rp. 1,084 trillion.The average value of capital goods for production in all large and medium industries in that year was worth Rp.54 billion.The average ratio of the ICT proportion to the total capital goods value was 0.003.The utilization value of non-ICT capital goods represented by fuel and lubricants had an average value of Rp. 3 billion.[30] [31] [32] The average total wage value for large and medium industrial workers in Indonesia in 2017 was Rp. 42 million per person annum.The average value of raw materials and supporting materials for large and medium industrial companies in Indonesia was Rp. 74 billion in 2017.  2 shows that only 2,739 companies exported their products.Meanwhile, the remaining 30,838 companies only marketed their effects on the domestic market in 2017.Based on investment ownership, it can be seen that 3,170 large and medium industrial companies in Indonesia are foreign investment companies, while 30,407 companies are domestic investment companies.Based on the company's size, in 2017, it was recorded that 11,000 companies are belonging to the large company group (having a workforce of 100 people or more) in Indonesia.As many as 22,577 companies are included in the medium-sized company (having a workforce of 20 -99 people).In terms of intensity of capital goods and labor, capital-intensive industries were as many as 29,542 companies, while labor-intensive companies are recorded at 4,035 companies.3 reports the estimation results regarding digitalization with labor productivity in large and medium industries based on Equation 1.More detailed information regarding the estimation results is shown in Appendix B. Statistically; all independent variables are simultaneously associated (significant) with labor productivity of large and medium industrial companies in Indonesia in 2017.Table 3 also shows that digitalization is robustly and positively associated with labor productivity of large and medium industrial companies in Indonesia in 2017.These findings are based on all six models' variable coefficient values, even though there are slight differences among each model.These results follow the hypothesis and prove that labor productivity can be improved through digitalization in Indonesia.Following the production theory concepts, technological improvements will impact output production capacity (positive) and cost savings.Another reason that supports these findings may be that the implementation of ICT will drive efficiency improvements in production time, energy, and materials inputs.The increase in labor productivity due to digitalization in this study can also be supported by the factor generated from human resources competence.With the implementation of ICT, the labor role becomes more central than conventional routine technical jobs.For example, workers are more needed to manage or troubleshoot machines' connectivity than to operate them by following procedures.As time goes by, the workers' managerial and problem-solving abilities will increase naturally, so that it will have implications in terms of productivity.These benefits are intangibly obtained by workers due to digitization.[42] Besides, the implementation of digitalization is correlated with sensor technology, connectivity, automation, accelerated information exchange, and data analysis, which allows the company to increase its production flexibility.[43] [2] Increased flexibility allows the company to adapt quickly to consumer demand variations to obtain the highest consumer surplus, as stated in The Production Theory [44] As a consequence, both company income and labor productivity will increase due to flexibility improvements.The estimation results in this study support Doms et al. (2004), Atrostic and Nguyen (2005), Biagi and Parisi (2012), Hall et al. (2012), and Kroll (2018).[10] [15] [27] [24] [45] They also found that the implementation of digitalization is positively associated with labor productivity in the industrial sector.These findings also support a study conducted by Khanna and Sharma (2018, p. 34).[14] They found that companies with low ICT penetration had lower labor productivity than those with high ones in India.In that study, the threshold for the high -low penetration of ICT was also determined by the proportion of ICT assets (the ratio of ICT assets to the total value of capital assets).On the other hand, this study's results showed contradictory results with Ito and Lechevalier (2009, [3] [10][11] [12] [14][15] [16] [24][25] ICT represents the ratio of ICT asset value to total capital value (unit ratio).[10][26][27] In terms of control variables, CAP represents the utilization of non-ICT capital[28] [29], which is measured by the logarithmic value of fuel and lubricant consumptions (million rupiahs).[30][31][32] WAG describes the logarithmic value of the annual average wage of labor (million rupiahs per person per annum).[3][16][33] MAT describes the logarithmic value of materials and consumables consumptions (million rupiahs).[34][35][36] [40] [15][41]

Table 1 .
Descriptive statistics of quantitative variables ISSN:

Table 2 .
Descriptive statistics of qualitative variables