The impact of technology adoption on employment structures and labor productivity a case study in vietnam manufacturing firms form 2007 to 2013

UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIESVIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS THE IMPACT OF TECHNOLOGY ADOPTION ON EMPLOYMENT STRUCTURES AND LABOR P

UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES

VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

THE IMPACT OF TECHNOLOGY ADOPTION ON EMPLOYMENT STRUCTURES AND LABOR PRODUCTIVITY A CASE STUDY IN VIETNAM MANUFACTURING FIRMS FROM 2007 TO 2013

BY

NGUYEN HUONG NGUYEN

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

HO CHI MINH CITY, NOVEMBER 2015

UNIVERSITY OF ECONOMICS INSTITUTE OF SOCIAL STUDIES

VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

THE IMPACT OF TECHNOLOGY ADOPTION ON EMPLOYMENT STRUCTURES AND LABOR PRODUCTIVITY A CASE STUDY IN VIETNAM MANUFACTURING FIRMS FROM 2007 TO 2013

A thesis submitted in partial fulfilment of the requirements for the degree of

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

By

NGUYEN HUONG NGUYEN

Academic Supervisor:

PROF.DR NGUYEN TRONG HOAI

HO CHI MINH CITY, NOVEMBER 2015

“The thesis entitled “The impact of technology adoption on employmentstructures and labor productivity A case study in Vietnam manufacturing firmsform 2007-2013” is the requirement for the degree of Master of Art in DevelopmentEconomics to the Vietnam – The Netherland Programme (VNP).”

Nguyen Huong Nguyen

I have taken efforts in writing this thesis However, it would not have beenpossible without the kind support of many people I would like to extend my sincerethanks to all of them

I wish to express my deep sense of gratitude and thanks to my supervisor,Prof.Dr Nguyen Trong Hoai for his invaluable guidance and constantencouragement which has sustained my efforts at all the stages of this thesis work.Sincere thanks also go to Dr Pham Khanh Nam for his valuable comments andsuggestions for my concept note as well as for his enthusiasm of helping mecollecting data It is also my duty to record my gratefulness to all VNP lecturerswho have helped and taught me a great deal of useful knowledge

My appreciations go to my classmates who have willingly helped me outwith their abilities and accompanied with me at VNP during two years I am alsovery much thankful to VNP officers and librarian for all their assistances ofavailable lab room, library and study materials

Finally, I owe a debt of gratitude to my parents for inspiring me and keeping

me going to this work

Nguyen Huong Nguyen

: Research and Development: Danish International Development Agency: Foreign Direct Investment

: State-Owned Enterprises: General Statistics Office: Central Institute for Economics Management: Gross Domestic Product

: Asian Productivity Organization: Human Identification Solutions Conference: Vietnam Rubber Association

: Ministry of Industry and Trade

Keywords: Technology adoption; employment structures; labor productivity;

operated personal computer; factory equipment; manufacturing firms

Table of contents

Chapter 1: Introduction 11

1.1 Problem Statement 11

1.2 Research objectives: 13

1.3 Research questions 13

1.4 The scope of the study 13

1.5 The structure of the study 13

Chapter 2: Literature review 15

2.1 Key concepts: 15

2.1.1 Technology and technology adoption 15

2.1.2 Employment structures 16

2.1.3 Labor productivity 16

2.2 The relationship between technology adoption and employment structures 17

2.3 The relationship between technology adoption and labor productivity 19

2.4 The relationship between firm characteristics and employment structures 21

2.5 The relationship between firm characteristics and labor productivity 22

2.6 The relationship between capital-labor ratio and labor productivity as well as the correlation of capital-labor ratio and employment structures 22

2.7 Conceptual framework: 23

Chapter 3: Data and Research methodology 24

3.1 Model specification 24

3.3 Data source 29

3.4 Estimate methods 31

Chapter 4: Technological revolution, employment structures and labor productivity in Vietnam manufacturing firms 32

4.1 Technology innovation 32

4.2 Employment structures 35

4.3 Employee productivity 36

4.4 Summary of the chapter: 38

Chapter 5: Empirical Results 39

5.1 Descriptive analysis 39

5.2 Empirical results 45

5.2.1 For the sample 45

5.2.2 For the specified industries 52

5.3 Summary of the chapter 61

6 Conclusion and policy recommendation 62

6.1 Conclusion 62

6.2 Policy recommendations 63

6.3 Research limitations 64

References 65

Appendix 69

List of tables

Table 3.1: Employment structures in SMEs 28

Table 3.2: Definition of variables 29

Table 4.1: Technological content of manufactured exports (%, 2000, 2008) 33

Table 4.5: Labor productivity by firm size and location 37

Table 5.1: List of considered industries 39

Table 5.4: The summary of statistics by mean of each industry 43

Table 5.5: The coefficient signs between employment structures and other independent variables 46

Table 5.7: The coefficient signs between the proportion of professional workers and other independent variables 49

Table 5.8: The coefficient signs between the proportion of sales and office workers and other independent variables 50

Table 5.9: The coefficient signs between the labor productivity and other independent variables 52

Table 5.10: The coefficient signs among the employment structures, labor productivity and other independent variables in the manufacture of food 53

Table 5.11: The coefficient signs among the employment structures, labor productivity and other independent variables in the manufacture of textile 55

Table 5.12: The coefficient signs among the employment structures, labor productivity and other independent variables in the manufacture of wood and of products of wood and cork, except furniture; manufacture of articles of straw and plaiting materials 56

Table 5.13: The coefficient signs among the employment structures, labor productivity and other independent variables in the manufacture of rubber and plastics products 58

Table 5.14: The coefficient signs among the employment structures, labor productivity and other independent variables in the manufacture of fabricated metal products, except machinery and equipment 59

Table 5.15: The coefficient signs among the employment structures, labor

productivity and other independent variables in the manufacture of furniture 60

List of figures Figure 3.1: The structure of industries considered 30

Figure 4.1: The proportion of enterprises that obtained a new technology by location and size 34

Figure 4.2: Levels of labor productivity per hour worked, 1970-2010 36

Figure 5.1: Changing in employment structures form 2007-2013 40

Figure 5.2: Added value per worker 41

Figure 5.3: Numbers of machineries and computer used 42

List of appendix Appendix 1-Table 4.2: Technology Characteristics (percent) 69

Appendix 2-Table 4.3: Worker Composition by Occupation (%) 70

Appendix 3-Table 4.4: Labor productivity by Sector from 2009-2013 71

Appendix 4-Table 5.2: Description of variables 72

Appendix 5-Table 5.3: Description of variables 74

Appendix 6-Correlation Matrix of variables 76

Appendix 7- Harris-Tzavalis unit root test 78

Appendix 8- Testing multicolinearity among variables by using VIF 80

Appendix 9-Testing multicolinearity among variables without BOTH variable by using VIF test 82

Appendix 10- List of industries in SMEs 85

Chapter 1: Introduction

1.1 Problem Statement

In many advanced countries over the past several decades, technology hasbecome an important part in most workplaces According to the United State’sBureau of Labor Statistics research in 1987, the rapid development and widespread

of technology is increasing in many industries, which helps manufacturers reducecosts and compete more effectively in both domestic and overseas markets Alongwith the wave of technologies, plenty of studies have already shown empiricalevidence that the introduction of advanced technologies in manufacturing haschanged the structures of employment and improved the labor productivity Forexample, Eli Berman, John Bound, Zvi Griliches (1993) and Doms, Dunne and R.Troske (1997) by assuming that nonproduction workers represent for a more skilledgroup of workers, authors investigated that the requirement of effectively adaptingnew techniques from innovation gradually increases the demand for skilled workersand technology are mainly responsible for job loss among low-skilled employees.Moreover, researchers also argued that such use of technology helps workers getmore tasks done within a short time to gain in productivity

In Vietnam, the manufacturing sector plays a crucial role as the main drivingforce of economic and productivity growth This sector contributes the most inVietnam’s GDP, creates stable jobs as well as contributes to the foreign trade inboth export and import (Le and Harvie, 2010; Huong, 2014) In the situation theglobalization and international competition have become important and innovation

is the essence of competitiveness Especially, the establishment of ASEANEconomic Community in 2015, that requires Vietnam has to concentrate more oninvesting in advanced technology and rely more on productivity to increasecompetition in global markets (World Bank, 2014) Over more than three decadessince economic reforms and industrialization in 1986, the great deal of newtechnologies has been introduced into Vietnam (Thuy, 2009) With the advantages

of low-cost labor forces and many supports from government to foreigninvestments, Vietnam has attracted a variety of technologically advanced and highervalue-added manufacturers That means there is a relative demand for skilled-workers corresponding to the acceleration of technology upgrading (Autor, Katz andKrueger, 1998) Simultaneously, McCaig and Pavcnik (2013) stated that there hadbeen the striking movements of Vietnamese employment from agriculture sectortoward manufacturing and services sectors Considerably, manufacturing sectorexperienced a rapid growth in labor productivity and a large employment expansion

in workforces Therefore, with the same spirit to Doms, Dunne and R Troske(1997) and Liu, Tsou and Hammett (2000), this study aims to investigate whethertechnology adoption affects on employment structure between non-productionworkers and production workers as well as whether technology adoption couldimprove the labor productivity in Vietnamese manufacturing firms from 2007 to

2013 However, almost studies used cross-sectional analysis to explore the impact

of advanced technology on nonproduction worker’s share and labor productivity inmanufacturing firms Doms, Dunne and R Troske (1997) also showed the results oftime-series analysis but they found it insignificant This study tries using panel data

to estimates the correlation between technology adoption and manufacturingemployment structures as well as labor productivity in Vietnam, the issue seems todraw little attention in developing countries involving Vietnam Moreover,understanding the effect of technology adoption on employment structures and laborproductivity could help Vietnamese policymakers propose various suitablestrategies to enhance the quality of labor and labor productivity, anticipatedrawbacks of technology progress on manufacturing labor and increase the nationalcompetitiveness in the international market

1.2 Research objectives:

The purpose of this paper is to reach the two following research objectives:(1) Investigate the effect of technology adoption on employment structure betweennon-production workers and production workers in Vietnamese manufacturingenterprises

(2) Investigate the effect of technology adoption on labor productivity in

Vietnamese manufacturing firms

1.3 Research questions

The study is to investigate the impact of technology adoption on employmentstructures and productivity in Vietnamese manufacturing firms The purpose of thispaper is to address two main research questions:

(1) Do technological advanced companies have a larger share of nonproduction workers?

(2) Do technological advanced companies gain higher productivity?

1.4 The scope of the study

The study will examine the influences of technology adoption onemployment structures and productivity in term of Vietnamese manufacturingenterprises The firm panel data used in this paper is collected from the Survey ofSmall and Medium Scale Manufacturing Enterprises (SMEs) in Vietnam from 2007

to 2013 These firms are classified according to the two-digit categories of TheInternational Standard Industrial Classification of All Economic Activity (ISIC)4from 10 to 33

1.5 The structure of the study

The research is organized as follow Chapter 1 is a chapter of introduction.Chapter 2 briefly reviews the existing theoretical and empirical literature onrelationship between technology and employment structures and the impact oftechnology on labor productivity Chapter 3 describes the research methodology

Next, chapter 4 presents an overview of technology, employment structures andproductivity in Vietnam In chapter 5, data and descriptive analysis, estimatetechnique and regression results of the analysis are presented Finally, theconclusion, limitations and some policy implications follow in chapter 6.

Chapter 2: Literature review

This chapter includes seven sections The first section is introduced howsome key concepts of the study involving technology, technology adoption,employment structures and labor productivity are defined The next six sectionsrepresent the literature review of the impact of technology adoption as well as firmcharacteristics on employment structures and labor productivity Then, theconceptual framework is illustrated in the last section

2.1 Key concepts:

2.1.1 Technology and technology adoption

Technology is reported as the state of knowledge concerning ways ofconverting resources into outputs According to Bartel and Sicherman (1999), themeasures of technology involve total factor productivity (TFP), the NBER TFPgrowth series, the ratio of investment in computers to total investment, the ratio ofR&D fund to net sales, the number of patents used in the industry and the ratio ofscientific and engineering employment to total employment Dunne and Schmitz(1995) and Doms, Dunne, and Roberts (1995) based on the type of productionequipment utilized in plants to present for technology variable In addition, Berman

et al (1994) used computer investment variable as a proxy for the rate of technologychange In the words of Hall and Khan (2002), the term of technology adoptionrefers to the decisions to acquire and utilize a new invention or innovation, whichresults from a comparison of the uncertain benefits of the new inventions with theuncertain costs of using them Meanwhile, Rogers (1983) supposed that technologyadoption is the process, in which an individual or an organization made a decision touse an innovation in their manufacturing Furthermore, Rogers claimed that beforeadopting a new technology, adopters would experience five stages includingawareness, interest, evaluation, trial, and finally adoption Also, in 1995, the authorexplored the adoption of technological innovations occurred not only within but alsooutside of organizations

2.1.2 Employment structures

Employment structure is represented as the shares of production workers andnonproduction workers (Liu, Tsou and Hammett, 2000) For instance,nonproduction workers, who are also referred to as white-collar workers, work inoccupations including manager, office workers, sales workers and professionals.Meanwhile, production workers, who are usually known as blue-collar workers ormanual workers, work in manufacturing activities such as fabrication and assembly,maintenance and repair equipments, material handling, warehousing and shippingproducts or security services Moreover, production workers also engaged inauxiliary production as well as other manufacturing related services Noticeably,apprentices are eliminated in term of production workers (OECD, Labor statistics2002)

2.1.3 Labor productivity

According to OECD, productivity is a ratio of an output volume and acorresponding input used (capital, land, raw materials, etc) In the broadest sense,Mukherji (1962) argued that productivity is the illustration of the use of resources

In other words, productivity is the effective combination of a variety of factors such

as scientific management, technology development, scientific allocation andutilization of resources and human Depending on what variable is chosen as themeasure of output and which input is concerned about, there are various kinds ofproductivity measures including capital productivity, multifactor productivity andlabor productivity However, labor productivity (an output per unit of labor input)plays a particularly important role in both the economic and statistical analysis of acountry (OECD) For instance, employee productivity presents the efficiency for allproductive activities Labors with higher productivity can produce more goods andservices than lower productivity labors with an equal number of work hours, andthen an economy is able to gain more for the same amount of work (U.S Bureau ofLabor Statistics)

2.2 The relationship between technology adoption and employment structures

The relationship between technology and employment structures has beenempirically investigated in several studies within many last decades Much evidencesupports the hypothesis that technology adoption has increased the share ofnonproduction workers in many workplaces

Ricardo (1821) examined the effect of the machinery adoption on thedifferent classes in the society The analysis pointed out that machinery displayedlabor The improvement in productivity as a result of mechanization reduced theproduction costs and obviously the real prices of goods While the landowning classand capitalists got more benefits from the lower prices, workers had to suffer athreat of losing jobs if capitalists reduce the wage fund to pay for the expensivemachinery, leading to technological unemployment among employees At this point,Ricardo showed that because of the competition among workers, wages were forceddown and the appearance of new machinery can lead to a drop in the well-being ofthe working class

Berman, Bound, and Griliches (1994) using published data from the U.SAnnual Survey of Manufactures stated that the defense buildup and trade deficitsare reasons for the slight shift in demand towards non-production workers.Meanwhile, production labor-saving technological change has the most likelyinfluence on the shift in demand towards non-production workers instead ofproduction workers Furthermore, this study reached the finding that industry-levelchanges in the nonproduction labor share are positively correlated with computerinvestment and R&D

Additionally, Timothy, Haltiwanger and Troske (1996) using plant-leveldata for U.S manufacturing from 1970s to 1980s found the relationship betweentechnological changes and the employment structure in U.S productive enterprises

In the research, authors concentrated on the share of nonproduction labor and therole of observable indicators of plant level utilized technology By assuming thatnonproduction labor is defined as skilled employees, the study concluded that

capital-skill and R&D-skill complementary have the positive connections with thesecular rise in the average share of nonproduction labor.

Doms, Dunne and R Troske (1997) explained the mixed findings when theyinvestigated the impact of technology on workers and wages at U.S manufacturingplants by using both cross-section and time series research The cross-sectionalresults indicated that more skilled workers were hired in plants which used a largenumber of advanced technologies In other words, more managers, professionalsand precision-craft workers were employed relatively in advanced firms.Accordingly, production workers were on average less skilled than nonproductionworkers and a positive change in the nonproduction labor share was seen asevidence of worker skill upgrading in an industry or a plant However, the timeseries research failed to prove that there is the correlation between technologyadoption and nonproduction labor share except for the result that plants adoptingnew factory automation technologies experienced the more skilled workforces bothpre-adoption and post-adoption Furthermore, the test results also revealed thatplants investing relatively more in computing equipment which was often main tool

of managerial and clerical labor experienced larger increase in share ofnonproduction worker

From the data of U.S manufacturing from 1909 to 1929, one of empiricalresults revealed by Goldin and Katz (1998) was that capital intensive and thecontribution of purchased electricity which created motive energy, led to theincrease in the number of educated production workers in plants

Using the survey of manufacturing firms, Liu, Tsou and Hammett (2000)asserted the impact of advance technology adoption on wage and employmentstructures in manufacturing firms of Taiwan With the similar model proposed byDunne and Schmitz (1995), the empirical results reported that firms using moreadvanced technologies hired a higher percentage of nonproduction labor, especiallyengineers, technicians, managers and supervisors

In the same way of using the nonproduction concept with the research ofDoms, Dunne and R Troske (1997), the paper of Dunne and Troske (2005) pointedout the connection between technology adoption and the skill mix of the labormarket in U.S productive plants Using the adoption of seven different informationtechnologies to measure the technology variable, the relationship betweentechnology adoption and workforce skill was found to vary across the technologies.For more details, the workforce that contained the large share of nonproductionemployees was associated with the use and adoption of engineering and designtasks Also, plants that adopted more technologies in terms of engineering anddesign tasks normally had the faster growth pace over the period 1987–1997 In thispaper, there was no evidence that proved the link between technology adoption andchanges in workforce skill at the plant level.

In summary, a general judgment is that almost previous studies provideevidences to support the hypothesis that firms adopting advanced technologies inproduction hire relative more fractions of nonproduction workers

2.3 The relationship between technology adoption and labor productivity

From the production function, Mankiw (2010) described the relationshipbetween technology and labor productivity through the following function:

Y/L = AF(1, K/L, H/L, N/L)

Where:

Y/L denotes the output per worker, which is a measure of productivity perworker;

K/L refers to physical capital per worker;

H/L refers to human capital per worker;

N/L represents for natural resources per worker

In this equation, physical capitals per worker, human capital per worker and natural resources per worker simultaneously have the effects on worker’s

productivity Moreover, productivity also depends on the technical knowledge,which is reflected by the variable A.

Besides, lots of empirical studies find out the positive correlation betweentechnology and labor productivity Lakhani (1982) using the time series as well ascross-section data of U.S coal mines in the Energy Information Administration for

1977, showed that adoption of the latest technologies increased labor productivity inboth underground and surface mines

In 1997, Black and Lynch estimated the Cobb Douglas production functionwith both cross-section and panel data in the period from 1987 to 1993 to examinethe effect of workplace practices, information technology and human capitalinvestments on productivity The authors figured out that the rate of managerialworkers who used a computer at work did not impact on labor productivity but theproportion of those who were non-managerial workers using computer at workaffect greatly to plant productivity At the same time, Doms, Dunne and R Troskesupported the same point of view that technologically advanced plant gained a highproductivity with cross-sectional analysis

Similarly, the OECD 1998 Technology, Productivity and Job Creation reportshowed evidence on the role of technology in economic performance The reportemphasized that due to the diffusion and adoption, technology directly improved theproductivity of innovating firms as well as indirectly raised economy-wideproductivity

Mcguckin, Streitwieser and Doms (1996) documented the correlationbetween the adoption of advanced technologies and productivity as well asproductivity growth rates from the 1993 and 1988 Survey of ManufacturingTechnology The main finding of the study was that enterprises that used advancedtechnologies gained higher productivity In addition, while the use of advancedtechnologies were able to improve productivity performance, their analysissuggested that the cross-section relationship showed the evidence that goodperformers were more likely to adopt advanced technologies than poorly performingestablishments

Huergo and Jaumandreu (2004) concentrated on the relationship betweentotal factor productivity growth and the introduction of innovation in Spanishproductive firms from 1990 to 1998 By measuring productivity growth by means ofthe Solow residual and using semiparametric methods, they concluded that processinnovations leaded to extra productivity growth at any point in this process Thisextra productivity persisted in rising for a number of years However, whenever theinnovation stops, all production growth in following years seems be disappearing.

Utilizing the conditional frontier approach, Filippetti and Peyrache (2012)examined the relative contribution of capital accumulation, exogenous technicalchange and efficiency, and as well as endogenous technological capabilities to laborproductivity growth in 211 European regions in 18 countries in the period 1995-

2007 They argued that the capital accumulation and exogenous technical changeare reasons for the convergence in labor productivity growth Nevertheless,advanced regions and backward regions had different relative contributions For thelatter, capital accumulation is the main cause which driven the productivity growth.Meanwhile, the convergence process faced some doubts because of the lack ofendogenous technological capabilities

All above empirical studies drive the hypothesis that technological advancedfirms experience higher labor productivity The differences among these researcheslie on the research methodology, specific country characteristics, researching periodand proxy variables

2.4 The relationship between firm characteristics and employment structures

The firm characteristics in this study will involve firm size, firm age and theshare of male workers in workforce

Liu, Tsou and Hammett (2000) explored that large firms employed a smallershare of managers and supervisors when they investigated the occupational mix ofworkers in plants of Taiwan

Manuel Adelino and Song Ma (2014) used regional industrial structure andnational changes in manufacturing employment also found that startups created jobs

because they were the source of new investment opportunities through innovation.

In contrast, Liu, Tsou and Hammett (2000) found that the rate of managers,supervisors, clerical and sales workers were higher among older firms

Wootton (1997) indicated that women and men had different trends inchoosing occupation For instance, women tended to engage in clerical and servicesoccupations while men highly concentrated on craft, operator and laborer jobs

2.5 The relationship between firm characteristics and labor productivity

Empitically, Leung, Meh and Terajima (2008) and Tran Xuan Huong (2014)found a positive relationship between firm size and labor productivity as well asTFP in both the manufacturing and nonmanufacturing sectors Particularly, theauthors concluded that the labor productivity relationship was even stronger in themanufacturing sector than the non-manufacturing sector

Taking into consideration the role of firm age with productivity, Huergo andJaumandreu (2004) argued that entrant firms experienced high productivity growthand that tended to coverage on average to common growth rates However, somestudies supported the relationship between firm’s age and productivity as the study

of Celikkol (2003) By concentrating on the U.S food and kindred productsindustry, this paper suggested that older plants had higher productivity growth ratesthan younger plants

Petersen, Snartland and Milgrom (2000) showed the evidence that womenwere slightly less productive than men in these typically male-dominated blue-collar occupations when they compared male and female workers working in thesame occupation in firm in three countries Sweden, U.S and Norway

2.6 The relationship between capital-labor ratio and labor productivity as well

as the correlation of capital-labor ratio and employment structures

Doms, Dunne and R.Troske, (1997) and Liu, Tsou and Hammett, (2000)pointed out that capital-labor ratio had the positive and significant with bothemployment structures and labor productivity For instance, capital-intensive firmshired a larger share of nonproduction workers and gained higher productivity

2.7 Conceptual framework:

The main purpose of this study is to examine the effects of technology adoption

on employment structures and labor productivity However, firm’s characteristicswhich are concerned as one of factors could lead to the changes in employmentstructures and productivity and firm’s characteristics will also be included In which,factory equipments and operated personal computer proxy for technology adoption Asmentioned previously, Vietnamese manufacturing firms usually apply two types ofmachineries intro production such as manually operated machineries (MOM) andpower driven machineries (PDM) Therefore, these types of machineries will be used toproxy for technology adoption Furthermore, the characteristics of firm which arefinally considered in researching are firm size (SIZE), firm age (FIAGE), the log ofcapital-labor ratio (CLR) and the proportion of male workers (MALE) Basing on theliterature review, the conceptual framework of the impact of technology adoption onemployment structures and labor productivity is presented as the following figure

Technology adoption

Factory equipments

Manually operated machinery only (MOM)

Power driven machinery only (PDM)

Both manually operated machinery

and power driven machinery (BOTH)

Operating personal computer (OPC)

Firm’s characteristics

Firm size (SIZE)

Firm age (FIAGE)

The log of Capital-Labor ratio (CLR)

Male ratio (MALE)

Employment structures

Labor productivity

Chapter 3: Data and Research methodology

The research methodology chapter contains four parts In the first part, theemployment structures and labor productivity models are illustrated Then, thesecond part shows the expected relationship between independent variables anddependent variables The next part introduces the data source, sample and variableconstruction Finally, the estimator methods are discussed in the last part of thechapter

3.1 Model specification

In this study, two models will be estimated The first is employmentstructures model which exams the change of employment structures when firmsapplied technologies into production The second is labor productivity model, whichinvestigates whether technology adoption helps improving the productivity of labor.This model will be driven from the Solow’s production function All these twomodels treat technology adoption variable as an exogenous variable Moreover, aseach industry and each occupation have different characteristics that lead todifferent responses to the effect of technology adoption Therefore, the researchmodels will be separately applied for each industry and different types of labors toexamine the impact of technology adoption on the changes of employmentstructures and labor productivity

Employment structures modeling

To investigate how technology adoption affects the composition ofworkforce, the estimated model is quite similar to the models proposed by Doms,Dunne and R.Troske (1997) and Liu, Tsou and Hammett (2000)

y it = f(TECH it , X it ) + µ it

Where:

yitz stands for a share of nonproduction workers including manager,professional workers, offices and sales workers The share of nonproduction

workers is defined as the ratio of a number of nonproduction workers to regular

labor force of firm Simultaneously, the proportion of manager, professionals,

officer and sales workers are respectively defined as the rate of a number of

manager, professionals, officer and sales workers to regular labor force

TECHit denotes the technology adoption in firm Empirically, technology

adoption could lead to the increasing of nonproduction workers in many workforces

(Ricardo, 1821; Berman, Bound, and Griliches, 1994; Timothy, Haltiwanger and

Troske, 1996)

Xit represents firm characteristics including firm size, firm age the

capital-labor ratio and proportion of male workers In some previous researches, the large

firms tended to hire less nonproduction workers (Tsou and Hammett, 2000) The

startups were supposed as the source of creating jobs (Manuel Adelino and Song

Ma, 2014) The capital-labor ratio seemed to have the positive correlated with the

share of nonproduction employees (Tsou and Hammett, 2000) In addition, in

manufacturing firms women tended to work in occupations like clerical workers or

sales workers (Petersen, Snartland and Milgrom, 2000)

And µit is the error term.

Labor productivity modeling

A production function is a mathematical representation, from which the firm

can choose to set up for its production process It shows the highest level of output

that a firm can produce for every specified combination of inputs

Where:

Y denotes the quantity of output,

K is the quantity of physical capital such as plant and equipment which used

in production,

L is the quantity of labor,

And A is a level of technology (TECH)

The average product of labor in the workforce (APL): APL = (2)

From equation (1) and (2) the labor productivity function can be rewritten as

following:

APL = Y/L =

APL represents the labor productivity on average

Taking into account the control variable as firm characteristics, the model

estimating the labor productivity in this study will be as following:

APL it = (Y/L) it = A it F(K/L, 1) it + X’ it + µ it

Where:

APLit stands for labor productivity which is also defined as value-added per

worker According to the definition in SMEs, labor productivity can be measured by

either real revenue per full-time employees or real value added per full-time

employee In this paper, labor productivity is measured as the value-added per

worker (Doms, Dunne and R Troske, 1997)

Ait is level of technology,

K/L is the capital-labor ratio (CLR),

X’it refers to firm characteristics (firm size, firm age and male

ratio), And µit is the error term.

As the results from papers of some economists such as Black and Lynch

(1997), Mcguckin, Streitwieser and Doms (1996), Huergo and Jaumandreu (2004),

adopting new technology in manufacturing process could boost the productivity of

workers Moreover, workers in capital-intensive firms produced more products than

others (Doms, Dunne and R.Troske, 1997)

With respect of additional control variables, in manufacturing sector, entrant

firms and firms with larger size experienced the higher labor productivity (Leung,

Meh and Terajima, 2008; Tran Xuan Huong, 2014; Huergo and Jaumandreu, 2004)

Furthermore, male workers tended to gain the higher productivity than female

workers (Petersen, Snartland and Milgrom, 2000)

This paper employs two measures of adopted technology including:

(1) A set of dummy variables representing for different types of

manufacturing machinery used;

(2) A number of operating personal computer (OPC) used

The set of technological dummy variables includes only hand tools, nomachinery (HT), manually operated machinery only (MOM), power drivenmachinery only (PDM), both manually and power driven machinery (BOTH) One

of the purposes of this paper is to compare the changes of employment structuresand productivity of firm when firm does adopt the technology and does not.Therefore, the dummy variable “only hand tools, no machinery” is also considered.However, this dummy variable will be omitted in the regression

In additional, the service workers as the classification of the Survey of Smalland Medium Scale Manufacturing Enterprises are cleaners, food preparer andservers The study assumes that both factory machineries and operating personalcomputer would not influence significantly to their jobs and productivities;therefore although service workers are still considered as nonproduction workers,this occupation will not be analyzed In the different way from most previousstudies, this paper will not use the concepts of skilled-workers and unskilledworkers to refer to nonproduction workers and production workers, respectively.The reason is due to the classification in the Survey of Small and Medium ScaleManufacturing Enterprises (SMEs) The table below shows how employmentstructures in Vietnamese manufacturing firms are defined in this survey Only a part

of production workers is unskilled workers In the survey, the term of unskilledworkers is called “Labor” Other occupations are still able to be included in amount

of skilled workers To control for industry-specific factors that influenceemployment structures and productivity in Vietnam manufacturing firms, this study willseparately run regressions for different industries rely on the two-digit classification

Table 3.1: Employment structures in SMEs

1 Managers (Top management) 6 Production workers

2.Professionals (university and 6.1 Foreman and supervisor

2.1.Engineer and similar 6.3 Mach maintenance/repair2.2.Accountant/Economist 6.4 Mach operator/assembler

4 Sales personnel

5 Service workers (cleaners, food prep/servers)

Source: The Survey of Small and Medium Scale Manufacturing Enterprises (SMEs).

Table 3.2: Definition of variables

Variables Definition of variables

Nonproduction The proportion of nonproduction workers including

managers, professionals, sales workers, office workers

Professionals The proportion of professionals

Sales and Offices The proportion of sales workers and office workers

FIAGE The age of firm since establishment (year)

CLR The log ratio of the book value of fixed capital stock to

number of regular labor

3.3 Data source

While there have been many studies on the effect of technological adoption

on employment structures and productivity of firms using cross-section data, untilrecently there has been very little direct research examine this impact with paneldata The panel data utilized in the study come from the Survey of Small andMedium Scale Manufacturing Enterprises (SMEs) in Vietnam from 2007 to 2013.This survey has been conducted by the cooperation of the Institute of Labor Studies,Social Affairs (ILSSA) in the Ministry of Labor, Invalids and Social Affairs(MOLISA) and Department of Economics, University of Copenhagen with fundingfrom DANIDA in three major cities (Ha Noi, Ho Chi Minh City, Hai Phong) and

seven provinces of Vietnam (Ha Tay, Phu Tho, Nghe An, Quang Nam, Khanh Hoa,Lam Dong, Long An) After excluding firms which are not interviewed in all fourgiven years or those collapsed in the period, the final data contains almost 1350manufacturing firms for analysis These firms are classified according to the two-digit categories of The International Standard Industrial Classification of AllEconomic Activity (ISIC)4 from 10-33 The samples of industries in SMEs are veryunequal Only 11 of total 24 industries have more than 30 surveyed firms The fivelargest samples belong to Manufacture of food products (10), Manufacture offabricated metal products, except machinery and equipment (25), Manufacture offurniture (31), Manufacture of wood and of products of wood and cork, exceptfurniture; manufacture of articles of straw and plaiting materials (16), andManufacture of rubber and plastics products (22).However, as the industry oftextiles is also an important industry of Vietnam Therefore, the Manufacture oftextiles (13) will be also considered in analyzing.

Figure 3.1: The structure of industries considered

Source: Author’s calculation from the data

3.4 Estimate methods

As mentioned previously, panel data which has more advantages than timeseries and cross-sectional data will be employed in the regressions Panel dataregression models include Fixed-Effects Model (Least Squares Dummy VariableModel); Random-Effects Models (Random Intercept or Partial Pooling Model) andPooled Ordinary Least Squares model (or Population-averaged Model) On the onehand, in order to select the most appropriate models for specified samples, Hausmantest, F-test as well as Breusch and Pagan Lagrangian multiplier test will be utilized.After figuring out the appropriate models, the robust regression will be employed tocontrol for heteroskedasticity and autocorrelation On the other hand, the varianceinflation factor (VIF) will also be used after regression models to detect themulticollinearity among the variables

Finally, as the data set is assumed that the number of panels tends to infinitywhile the number of time periods is fixed Therefore, the Harris-Tzavalis test, which

is introduced by Harris–Tzavalis (1999), will be used to test the stationary of thedata All tests are displayed in the appendix of this paper

Chapter 4: Technological revolution, employment structures and labor

productivity in Vietnam manufacturing firms

Before testing the impact of technology adoption on employment structuresand labor productivity, this chapter provides the overview about the performances

of technology revolution, employment structures and labor productivity in Vietnammanufacturing firms through four sections The first three sections describe theperformances of technology innovation, employment structures and employeeproductivity, respectively Meanwhile, the fourth section indicates the summary ofthis chapter

4.1 Technology innovation

Statistically, technological innovations was absent in the growth of Vietnameconomy over the period from 1975 to 2005 (Ngoc, 2008) Recently, the innovationsystem in Vietnam has just emerged It is still nascent with weak science,technology and innovation capabilities (World Bank, 2014) Accordingly, Vietnamhas joined in global value chains in a plenty of sections involving textiles, garment,food and furniture Nevertheless, the pace of producing high technology exports hasbeen extremely slow (Anh, Hung, Mai, 2013)

Table 4.1 shows the structure of technologies used in manufacturing exports

of Vietnam compared with other countries in two year 2000 and 2008 In which,while rations of medium and low technology had increased, that of hightechnologies slightly reduced from 11,1% in 2000 to 10,1% in 2008 In this group

of countries, Vietnam was one of the three nations which have the lowestpercentages of high tech exports over the period The adoption of high technologiesinto manufacturing in Vietnam is only higher than Cambodian (0.1% in 2008) Thismeans that Vietnam is still left behind many countries in adopting new and hightechnologies in production One of the reasons of this problem is that Vietnamesemanufacturing firms tend to be more labor-intensive

Table 4.1: Technological content of manufactured exports (%, 2000, 2008)

Source: Vietnam competitiveness report 2000, 2008.

In the words of Viet, Hien, Quy and Qui (2011), although Vietnam

Government had proposed policy to encourage technology change but technology

condition was not significantly reformative In private and FDI sectors, the

technological change was still lower compared with other countries in the region

and in the whole world because of the lack of skilled workers who were able to

apply new and advanced technologies Therefore, high technology was likely to

only adopt in a few of Vietnamese enterprises Percentage of Vietnamese high

technology enterprises are only 2%, whereas, those of Thailand, Malaysia and

Singapore was 30%, 51% and 73%, respectively Besides, Vietnam reached a very

low rank of competitiveness, especially in technological indicators (World

Economic Forum, 2009) Accordingly, the Vietnam ranks of creative and innovation

indicator respectively were 55/133 and 99/133 Furthermore, the types of

equipments used in almost firms were manually operated machinery and power

driven machinery The percentages of hand tools also gradually decreased from

7.8% to 5.0% but most of utilized technologies were quite old A number ofmachinery belonged to a range of 6-20 years old accounted for large percentages(Table 4.2, Appendix 1)

Figure 4.1: The proportion of enterprises that obtained a new technology by location and size.

Source: The Survey of Small and Medium Scale Manufacturing Enterprises, 2007- 2013

Figure 4.1 reports enterprises that obtained a new technology by location andsize from 2007 to 2013 Overall, fewer enterprises adopted new technologies overthe period The rate of new technology introduced in the manufacture processdropped from approximately 15% in 2007 to 6.4% in 2013 The reason is explained

by the decline in innovation ratios, which often lead to adoption of newtechnologies In other words, the pressure of financial crisis resulted in a higherlevel of uncertainty in doing business leading to the limited demand for newtechnology in companies Noticeably, from 2007 to 2011 urban firms were morelikely to adopt new technologies compared to rural firms but this advantage wasdisappear in 2013 when the percentages of adopted technologies are nearly equal in

both areas Moreover, the side-effect seems to exist It indicated that largerenterprises were more likely to utilize new technologies in manufacture.

4.2 Employment structures

Before the reform in 1986, the labor market in Vietnam experienced strictregulations Managers or supervisors did not have any influence in SOEs on hiringemployees and setting wage as these issues were directly controlled by thegovernment Over the past three decades, there are a number of positive changes inVietnamese labor market conditions, which regards to changes in regulations onrecruiting and firing labor and wage policies

According to the Statistical Yearbook of GSO, the manufacturing sector had

a remarkable employment growth rate in the period 1990-2011, increased from2.8% to 7.3% after 20 years This growth contributed to create jobs which caused alarge transition in labor away from the agricultural sector Moreover, the report ofCIEM indicated that there were reverse fluctuations among occupations in thisperiod from 2007 to 2013 in Vietnamese manufacturing enterprises While someoccupations such as manager, professional and office experienced the gradually rise,the percentage of production worker declined significantly from 66.2% in 2007 to59.3% in 2013 (Table 4.3, Appendix 2) That seems to be a transition betweennonproduction workers and production workers over the period Noticeably,enterprises which had larger sizes or those were in urban areas tended to hire moreprofessional workers as large firms usually had enough financial ability to applynew technology into production, besides, the advantages of location in big citieshelp firms easy to approach advanced technology

4.3 Employee productivity

The productivity of Vietnamese manufacturing firms had improved over theperiod 2000-2010 (Tran Xuan Huong, 2014) However, levels of labor productivity

of Vietnam were still low compared with international areas In the period from

2011 to 2013, the labor productivity of Vietnam continued to increase from 5.08%

to 5440 USD/labor (convert to the fixed price 2005 PPP) while the productivity

growth of Vietnam in the period 2007-2013 was only 3.9% Compare with other

countries in areas, Vietnam labor productivity equaled to 1/8 labor productivity ofSingapore and equals to 1/3 Thailand productivity Now, productivity of Vietnamwas only higher than those of Myanmar and Cambodia and approximate to Lao(Ngoc and Thu, 2013) The reason could be as while neighbor countries includingSingapore and Thailand’s workers concentrate on high added value sector likeservices, Vietnamese workers mainly focus on the textile and garment sectors whichcreate lower added value

Figure 4.2: Levels of labor productivity per hour worked, 1970-2010

Note: GDP at constant basic prices per hour, using 2005 PPPs, reference year 2010, USD Source: APO (2012), APO Productivity Data book 2012, Keio University Press, Tokyo

In Vietnam, a large number of the labor force still engage in agriculture andunofficial economic sectors which have lower productivity than manufacturing andservices sector (Table 4.4, Appendix 3) Therefore, they rarely have opportunities toapproach machines and modern technologies, leading to the low labor productivityrate In recent surveys, the International Labor Organization (ILO) concluded thatmanufacturing and services sectors gained much higher labor productivity thanagriculture sector They supposed that modern technologies and machines inproduction along with careful training strategies for workers could be the key toimprove labor productivity Besides, at the same time, the labor productivity ofmanufacturing firms gradually rose through the scale of firms Table 4.5 shows thelabor productivity by firm size and location For instance, firms with larger sizeshad higher productivity compared with other firms with smaller sizes The sametrend when compared the productivity of urban and rural locations.

Table 4.5: Labor productivity by firm size and location

4.4 Summary of the chapter:

Overall, in order to integrate successfully to the international market,Vietnam has to concentrate on investing and developing technologies, improvingthe quality of labor forces and relying more on productivity Noticeably, the results

of statistic showed that large firms and firms were in urban areas tended to adoptmore new technologies than others Simultaneously, CIEM concluded that therations of professionals, offices and sales workers in Vietnamese manufacturingfirms increased matching with the increasing in scale of firms and urban areas Thatleads to the issue that whether the technology adoption correlates with the share ofnonproduction workers in manufacturing enterprises or not? Furthermore,enterprises which had large size and located in urban location also achieved higherproductivity compared with smaller firms and those are in rural areas Inadditionally, there is evidence that firms which adopt technology experienced thestriking labor productivity than others To clarify, these issues will be tested anddiscussed in the next chapter

Chapter 5: Empirical Results

The empirical results are discussed within two parts The first part presentsthe data descriptive analysis which introduces the variables statistics and the initialcorrelation among variables In the next part, the regression results and discussionsare exposed

5.1 Descriptive analysis

As the results of data analysis, among industries, the manufacture of food(10) had the highest share of nonproduction workers, especially the proportion ofmanagers The manufacture of fabricated metal products, except machinery andequipment (25) and manufacture of furniture (31) followed as the second and thirdpositions The rations of nonproduction workers in other industries were quite similar.Overall, in all industries the percentages of professional workers, office workers and salesworkers were pretty scarce (Table 5.2, Appendix 4)

Table 5.1: List of considered industries

ISIC two-digit Classification Industry

13 The manufacture of textiles

16 The manufacture of wood and of products

of wood and cork, except furniture;manufacture of articles of straw and plaitingmaterials

22 The manufacture of rubber and plastics

products

25 The manufacture of fabricated metal

products, except machinery and equipment

31 The manufacture of furniture

Source: United Nations Statistics Division

Figure 5.1: Changing in employment structures form 2007-2013

Source: Author’s calculation from the data

The analysis of changing in the employment structures from 2007 to 2013indicates that there were the reverse trends in the shares of nonproduction workersand production workers in firms (Figure 5.1) While a large number of productionworkers dropped remarkably within the period, from approximately 8600employees to nearly 6730 employees, the share of nonproduction workers increasedslightly It consists with the results of performance analysis in chapter 4 (Table 4.3,Appendix 2)

In term of labor productivity, the manufacture of wood and of products ofwood and cork, except furniture; manufacture of articles of straw and plaitingmaterials (16) gained the lowest added value, only 38898.27 (1,000 VND), justnearly equaled to a half of manufacture of rubber and plastics products (22) (Table5.2, Appendix 4) However, when observing the changes of added values in eachindustry and total of them, the productivity of Vietnamese manufacturing firmssoared significantly, particularly the two industries involving the manufacture ofrubber and plastics products (22) and the manufacture of fabricated metal products,except machinery and equipment (25) In 2013, the added values of all industriesincreased over 2 times compared with the added value in 2007 (Figure 5.2)