Luận văn thạc sĩ UEH impact of rd on the productivity growth of manufacturing firms in vietnam

RATIONALE OF THE RESEARCH

In the modem economy today, technological progress has a quite central role It contributes importantly to growth of economy and is a key factor to determine the competitiveness of firms in both national and international marketplace Research and Development (R&D) is widely regarded as the core of technological advance, and innovative capacity of firms are reliably indicated by levels and rates of R&D expenditures growth Countries belonging to the Organization for Economic Cooperation and Development (OECD) spend significant amounts on R&D activities

On average, OECD countries have spent more than 2 percent of GDP on annual public and private R&D investments during the last two decades (OSTP 1 , 1997)

In a traditional way, firms have paid attention to R&D because the technical advances resulting from innovation may allow them to improve productivity, succeed in competitive markets, and meet environmental and regulatory requirements Besides, R&D has also had contribution to the development of new products and, in many cases, the creation of new markets Within firms, economic returns are always taken into consideration on deciding the importance and nature of R&D performance Firms usually take part in R&D activities only when the results are appropriate and offer higher rates of return than that of other available investment alternatives such as acquisition of new machinery, advertising, or purchase of speculative assets

There are many sources for productivity improvements, but one strategy for enhancing productivity growth which is widely acknowledged is increasing the stock of knowledge This stock of knowledge can be increased by formal investment in

1 OSTP is Office of Science and Technology Policy

R&D activities In the private and public sectors, the allocation of resources toward the investment to generate new knowledge must be decided carefully

In spite of the importance of R&D in firms' productivity, R&D activities have not been taken into consideration for much investment in Vietnam, especially in business sector While most OECD countries and China devoted around 2% of their GDP to R&D activities, Vietnam spent only 0.5% of its GDP for this purpose (Nguyen and Tran, n.d.) R&D expenditure of Vietnamese enterprises accounted for only about 20% of the total R&D expenditure of the country in 2002 (Nguyen, n.d.) Whereas, according to OSTP (1997), companies in OECD countries finance more than 50% of all R&D expenditure and they conduct two-thirds of all R&D activities SMEs make up the vast majority of registered companies in Vietnam, namely 96.5%

Nevertheless, the technology level across the SME sector in Vietnam is generally assessed as being two, three or even more times lower than both world and regional levels (Bezanson et al., 2000)

One of main reasons under the assessment of the Ministry of Industry is that the labor currently lack of necessary skills to support technological upgrading and there are very little R&D activities appropriate for such upgrading Indeed, only a small fraction of the country's R&D scientists and engineers are working in industrial enterprises The rest are working in national centers for R&D, ministries and government agencies, universities or other institutions that perform research Another reason is that there is little market-oriented relationship between firms, R&D institutions and universities (Bezanson et al., 2000) Moreover, the most important reason for a little investment in R&D activities of Vietnamese enterprises may be their limitations in financial resources

The case of Vietnam raises a doubt if R&D has any relationship with productivity of manufacturing firms Practically, there are many empirical studies at firm level that has emphasized the role of technological or knowledge capital in productivity growth

Early studies focused on R&D investment and found that in most countries, R&D has a significant contribution to productivity growth, especially in the cross sectional dimension However, the conclusion has not been verified in Vietnam.

OBJECTIVE OF THE RESEARCH

Research and development (R&D) investment has been regarded as an important factor in the improvement of productivity levels of firms This has been proved true by many empirical studies for many countries but neglected for Vietnamese case

Therefore, based on previous studies, the research is going to examine the relationship between R&D activities and productivity growth of manufacturing firms in Vietnam to answer the following questions:

Is there a positive impact of R&D on productivity growth in Vietnamese manufacturing firms?

What should those firms do to increase their productivities? and What policies should be recommended to support them m improving productivity by increasing R&D expenditure?

RESEARCH METHODOLOGY

The thesis studied the impacts of R&D expenditure to productivity growth of Vietnamese manufacturing firms ~y using data from the Vietnam Enterprise Survey

2004 The thesis used such methods as descriptive statistics, quantitative analysis and OLS regression to deal with the research questions.

THESIS STRUCTURE

The thesis consists of six chapters The first chapter is Introduction, which presents the rationale of the research, the objective of the research, research hypothesis as well as methodology, and the thesis structure The next one is Literature Review This chapter examines theories and empirical studies relating to the impact of R&D expenditure on productivity growth of manufacturing firms R&D activities of firms are discussed in the chapter 3: Overview of R&D and firm performance in Vietnam

Chapter 4: Research Methodology focuses on model specification and variables choices justification The practical results are analyzed via descriptive statistics and regression analysis in chapter 5: Result Analysis Finally, conclusions and policy recommendations are provided in Conclusions and Recommendations chapter.

LITERATURE REVIEW

INTRODUCTION

This chapter aims at reviewing literature related to the topic to make sure that the research is conducted based on a scientific background The chapter will be presented in three main parts In the part one, key concepts related to the topic such as R&D, productivity as well as manufacturing will be discussed Economic theories supporting for the study are found out and stated in the next part At the end of this part, a research model which represents factors affecting productivity is suggested

Finally, empirical studies regarding effect of R&D on productivity growth of manufacturing firms in some countries are discussed in the last part Through this chapter, impact of R&D expenditure on productivity growth is generally figured out on the basis of economic theories and empirical studies.

CONCEPTS

OECD (1994) defined that "Research and experimental development (R&D) comprise creative work undertaken on a systematic basis in order to increase the stock of knowledge, including knowledge of man, culture and society, and the use of this stock of knowledge to devise new applications" R&D has been divided into three categories: basic research, applied research and experimental development o Basic research is experimental or theoretical work that is undertaken not to obtain long-term benefits but to advance the state of knowledge (CBO, 2005) In basic research, characteristics, structures and relationships are analysed with a view to formulate and test hypotheses, theories and laws The results of basic research are not for sale but usually for publishing in scientific journals or usage of interested people Sometimes, it may be kept secret for security reasons o Applied research is also original work that is undertaken to acquire new knowledge with a specific application in view Its' aims are determining possible uses for results of basic research or determining new ways to achieve specific objectives Results of applied research are mainly valid for a limited number of products, operations, methods or systems The knowledge or information resulting from applied research is often applied for patent or may be kept secret o Experimental development is systematic work usmg existing knowledge gained from research and practical experience These research and experience is directed toward producing new materials, products and devices; installing new processes, systems or services; or substantially improving what has been produced or installed in the past

For example, basic research is the theoretical investigation of factors which have influence on regional variations in economic growth Applied research is the investigation that is performed for the development of government policy

Experimental development is the development of operational models based on laws with the purpose of modifying regional variations

"Expenditure on R&D may be made within the statistical unit or outside it" (OECD,

1994 ) The measurement of such two kinds of R&D expenditures is so complicated with many costs should be included or excluded However, in this thesis, R&D expenditure used to examine its effects on productivity growth of Vietnamese manufacturing firms is available in the Vietnam Enterprise Survey

Scientific and technological innovation is known as the transformation of an idea into a new or improved product, a new or improved operational process or a new approach toward a social service There are different meanings in different contexts for the word "innovation" depending on certain objectives of measurement or analysis New products or processes and significant technological changes in products or processes are considered as technological innovations An innovation is performed if it is brought out to the market or used in a production process Thus, innovations include dozens of activities relating to science, technology, organization, finance and commerce R&D is one of such activities and it may be done at different stages of the innovation process 2 • R&D can act as the origin of inventive ideas or a form of problem-solving (OECD, 1994) According to Rogers (1998), R&D is an important input measure of innovation

According to OECD (200 1 ), "productivity is commonly defined as a ratio of a volume measure of output to a volume measure of input use" This general concept has received no disagreement and can be applied in various ways That means there are many purposes and many ways to measure productivity The objectives of productivity measurement can be stated as follows:

A frequently stated objective of measuring productivity growth IS to trace technical change

Productivity growth is also measured to identify changes in efficiency which is conceptually different from technical change Full efficiency in an engineering sense means that a production process has achieved the maximum amount of output that is physically achievable with current technology, and given a fixed amount of inputs

A real way to describe the essence of measured productivity change IS to identify real cost savings in production

In the field of business economics, comparisons of productivity measures for specific production processes can help to identify inefficiencies

Measurement of productivity is a key element to assess the standard of living

2 See Appendix 1 for explanation of innovation process

There are many different ways to measure productivity, which depend on the purpose of productivity measurement or the availability of data Productivity measures can be divided into two kinds: single factor productivity measures and multifactor productivity measures Single factor productivity relates a measure of output to a single measure of output, whereas, multifactor productivity relates a measure of output to a bundle of inputs At the industry or firm level, there is a distinction between productivity measures that relate some measures of gross output to one or several inputs and those which use value-added to capture output movements

Table 2.1: Overview of main productivity measures

Type of Capital, labor and output

Labor Capital Capital and labor intermediate inputs measure (energy, materials, services)

Gross productivity productivity KLEMS multifactor output (based on gross (based on gross (based on gross productivity output) output) output)

Capital-labor MFP Value productivity productivity added (based on value (based on value (based on value added) added) added)

Single factor productivity measures Multifactor productivity (MFP) measures Source: OECD, 2001

In this thesis, gross-output based labour productivity, which is a ratio of quantity index of gross output to quantity index of labour input, is used to measure productivity Labour productivity is a useful measure because it relates to the most important factor ofproduction_labour and is relatively easy to measure

According to the US Census Bureau, the manufacturing sector includes establishments which are used in the physical or chemical transformation of materials, substances, or components into new products Except activities in the Construction sector, manufacturing is also considered as the assembling of parts of manufactured products, the blending of materials, and some other related activities

Manufacturing establishments are often known as plants, factories or mills They may process materials by themselves or sign contracts with others to process their materials for them Manufacturing establishments transform materials, substances or components which are raw products of agriculture, forestry, fishing, mining and so on The new products of manufacturing establishments may be finished products, which are ready for use or consumption, or semi-finished products, which become inputs for other establishments to use in further manufacturing

The manufacturing sector is divided into sub-sectors depending on different production processes with different kinds of material inputs, production equipment and employee skills In assembling activities, when parts and accessories of manufactured products are made for separate sale, they belong to the industry of the finished manufactured item For example, the manufacturing of replacement refrigerator door is classified in the refrigerators manufacturing However, the classification of components, which are input for other manufacturing establishments, is based on the production function of the component manufacturer For instance, electronic components belong to Computer and Electronic Product Manufacturing and stamps belong to Fabricated Metal Product Manufacturing.

ECONOMIC THEORIES

According to Pindyck and Rubinfeld (1992), the Cobb-Douglas production function is a widely-used approach to represent the relationship between an output and inputs in microeconomics Knut Wicksell proposed the function in the period 1851-1926,

- - - - - - - - and then in 1928 Paul Douglas and Charles Cob tested it against statistical evidence

The production function has the form as follows:

• Q denotes output, L: labor input, K: capital input

• A is a constant depending on the units in which inputs and output are measured

• a and p are the output elasticity of labor and capital, respectively These values are constants and ordinarily smaller than one because the fact that the marginal product of each input diminishes when that factor increases

Output elasticity measures the responsiveness of output to a change in levels of either labor or capital used in production, ceteris paribus For example, if a= 0.15, a 1% increase in labor would lead to approximately a 0.15% increase in output

Furthermore, if a + p = 1, the production function exhibits constant returns to scale If a + P < 1, there are decreasing returns to scale, and if a + p > 1, then there are increasing returns to scale For example, if L and K each are increased by 20%, Y increases by 20% when a + p = 1 Y increases more than and less than 20% when a +

P < 1 and a + p < 1, respectively The Cobb-Douglas production function is sometimes written in logarithmic form: log Q = log A + a log L + p log K This form is useful when performing a regression analysis

Pindyck and Rubinfeld (1992) stated that a general production function, Q = F(K, L), applies to a given technology This means a given state of knowledge might be used in the transformation of inputs into output When technology is improved and the production function changes, a firm can obtain more output with a given number of inputs For instance, a new and faster computer chip may enable a hardware manufacturer to produce computers with higher speed in a given period of time

The Cobb-Douglas production function helps to illustrate a way to measure production functions However, it is often replaced by other more complex production functions in industry studies for some reasons One of the reasons according to Pindyck and Rubinfeld (1992) is that the Cobb-Douglas function does not allow a possibility happening in the reality The possibility is that the firm's production process shows increasing returns at low output levels, constant returns at intennediate output, and decreasing returns at high output levels

2.3.1.2 The Law of Diminishing Returns

Pindyck and Rubinfeld (1992) stated the law of diminishing retums that "as the use of an input increases (with other input fixed), a point will eventually be reached at which the resulting additions to output decrease" When the labor input is small and capital input is fixed, a small increase in labor input will lead to a substantial increase in output because workers are allowed to develop specialized tasks However, when too many workers are used in the production, some of them become ineffective and therefore the marginal product of labor falls That is called the law of diminishing retums

The law of diminishing retums is often applied in short-run analyses because according to the definition, at least one input is fixed However, it sometimes can be applied to long-run analyses There is one point needed to pay attention to is that the law of diminishing retums differs from decrease in output due to changes in the quality of labor when labor input are increased For instance, when the most qualified workers are hired first, the output will increase much accordingly However, the output may not go up or go up at a low level when the least qualified workers are hired last In the analysis of production, we have to assume that the quality of all labor input are the same Diminishing retums result from limitations on the use of other fixed inputs such as machinery, not from declines in worker quality Moreover, we should not confuse diminishing retums with negative returns In the law of diminishing return, a declining marginal product is described, not a negative one

In this law, a given production technology is also assumed However, over time, inventions and technology improvements may allow the entire total product curve to shift upward, thus, more output can be obtained with the same inputs Although any given production process has diminishing returns to labor, labor productivity can increase if there are improvements in the technology According to the figure 2.1, improvements in technology may allow the output curve to shift upward from 0 1 to

Figure 2.1: The effect of technology improvement

Labor per time Source: Pindyck and Rubinfeld, 1992

According to Griliches (2000), the R&D capital model is still the most important research method today in estimating the effects of R&D on productivity growth, in spite of its many weaknesses It is a simple and easily-applied model that enables us to estimate the rate of return to R&D and then to measure its contribution to productivity growth Most of applied studies are based on it We can study many different forms of R&D capital such as private, public, and R&D done by neighboring firms or industries The first, direct approach IS represented by the equation as follows:

Y denotes some measures of output at the firm, industry, or national level;

X is a vector of standard economic inputs such as man-hours, structures and equipment, energy use, and so on;

K is one or more measures of cumulated research effort or "knowledge capital"; a(t) indicates other factors that affect output and change systematically over time; u reflects all other random fluctuations in output

This equation is taken in logarithmic form from the Cobb-Douglas production function It is the first approximation to represent a potentially much more complex relationship In this first equation, y, the elasticity of output with respect to research capital, is focused to be estimated R&D capital is often calculated by a weighted sum of past R&D expenditures with the weights reflecting both the potential delays in the impact of R&D on output and its possible eventual depreciation

In the second approach, growth rates are used to replace levels and the above equation becomes as follows:

The term y~log K is simplified as follows: p = dY/dK = y(YIK), ~log K = RIK, y~log K = RIK*p*(KIJ)

R is the net investment in K, net of the depreciation of the previously accumulated R&D capital;

- p is interpreted as the gross rate of return to investment m K, gross of depreciation and obsolescence;

In this form, the growth rate of output or productivity is related to the intensity (R/Y) of the investment in R&D or some more general measure of investment in science and technology

In the application of this model, there are a number of conceptual difficulties First, it is difficult to measure output and output growth accurately in science and technology sectors conceptually Second, the construction of R&D capital variable may also face issues of timing, depreciation and coverage and others The biggest problem with this model may be that it treats R&D and science as another kind of investment However, investing in the creation of knowledge is not similar to buying a machine or building a plant It is quite difficult to measure the results of such activities Nevertheless, this simple model is conveniently a starting point to examine empirical works in this area and applicable to our problem if we are able to consider their conceptual and data problems

With reference to econometric issue on applying this model, Griliches (2000) stated that there is simultaneity problem referring to possible confusion in causality: "future output and its profitability depend on past R&D, while R&D, in turn, depends on both past output and able to build a system of equations in which current output depends on past R&D, and past R&D depends on past output" However, with cross-sectional data, it is much more difficult to make such distinctions

2.3.3 Suggested research model from economic theories

Based on the above economic theories, the relationship between firm productivity and its determinants can be described in a function with dependent and independent variables as follows:

Y denotes measures of output of firms It is expressed in the form that representing labor productivity of firm

L denotes labor input of finn

K is physical capital of firm

R denotes measures of R&D capital CHAR is considered as some characteristics of firm which affect its productivity such as size of labor or type of ownership.

EMPIRICAL STUDIES

Being aware of the importance of research and development, many analysts have examined the relationship between R&D expenditure and productivity growth at firm level As a result, there is a great number of empirical studies estimating the impact of R&D investment on such growth According to CBO (2005), the results of such relationship spread a wide range Some researches have found that R&D virtually has no effect on productivity Whereas, other studies have discovered that R&D's effect is substantial and larger than effect of other kinds of investment However, most of the estimates lie somewhere between the two extremes, therefore, there is an agreement with the view that the relationship between R&D spending and productivity growth is positively significant

Mairesse and Sassenou (1991) conducted a research which surveys econometric studies examining the relationship between R&D and productivity at the firm level and assesses the results as well as problems encountered According to those authors, the Cobb-Douglas production function is the basic analytical framework used by most econometric studies that estimate the contribution of R&D on productivity growth In addition to such usual factors of production as labor, physical capital, materials and so on, a measure of R&D capital is also included in the function as explanatory variable

The Cobb-Douglas production function has an advantage that it can be estimated as a linear regression if all variables are transformed into logarithmic forms

On viewing problems encountered as mentioned above, Mairesse and Sassenou

( 1991) stated that econometricians try to simplify phenomena which are often highly complex ones This is especially true with R&D activities and their impacts on productivity They said that "R&D effects are intrinsically uncertain, they often happen with long lags, they may vary significantly from one firm or sector to another and change over time" The effects of other factors of productivity that happen simultaneously and have domination may make R&D effects to be hidden If serious problems in measuring variables and collecting good data are ignored, it is difficult to build up a production function between R&D and productivity Therefore, the authors were surprised to find out that in most studies, estimates of the R&D elasticity or R&D rate of return are statistically significant and frequently plausible

The above are what CBO (2005) and Mairesse and Sassenou ( 1991) found out when reviewing and synthesizing related studies However, the three case studies below will help to investigate further the relationship between R&D and productivity in practice

2.4.2 R&D and Productivity in French manufacturing firms

Cuneo and Mairesse (1983) investigated if there is a significant relationship between R&D expenditures and productivity performance at the firm level in French manufacturing industry for the period 1972 - 1977 The sample including 182 firms is divided into two sub-samples: scientific firms which belong to the R&D intensive industries such as chemicals, drugs, electronics and electrical equipment and other firms in other manufacturing industries The basic model used in this research is the simple extended Cobb-Douglas production function, which can be written in logarithmic form as follows:

Where i, t refer to the firm and the current year; e is the error tenn in the equation; v, c, 1 and k stand for production (value added), physical capital, labor, and R&D capital, respectively; Jl = a + ~ + y is the coefficient of returns to scale; and 'A is the rate of disembodied technical change

In this study, production is measured by deflated value-added V rather than by deflated sales Labour L is measured by the number of employees, physical capital stock C by gross-plant adjusted for inflation R&D capital stock K is calculated by the weighted sum of past R&D expenditure which use a constant rate of obsolescence of

15 percent per year Two variables, labor and physical capital stock are corrected for the double counting because they are already included in the R&D capital stock

Thus, the available number of R&D employees is simply subtracted from the total number of employees Whereas, the part of physical capital stock used in R&D is calculated based on the average ratio of the physical investment component of R&D expenditures to total R&D expenditures and is also subtracted However, in the practice of Vietnam, because having full financial statements of examined firms is very difficult, it is impossible to separate the part of physical capital in R&D expenditure from the total physical capital stock

The authors finally come up with discrepancies between the total and within-firm estimates of the two main parameters: the elasticity of physical capital stocks (a) and R&D capital stocks (y) However, due to good measures of the variables, the problem is much less serious than it could have been, and in general the estimates are statistically significant and likely high Besides, in order to find out further results, the authors used sales instead of value added and included and excluded materials M in tum in the production function The total estimates using sales and omitting materials do not differ much from those obtained with value added The within-firm estimates with sales instead of value added are also similar when constant returns to scale is imposed However, if constant returns to scale is not imposed, large discrepancies between the total and within-firm estimates occur The within-firm estimates are much improved when materials are taken into consideration Hence, the omission of materials in the sales specification affects especially the within-firms estimates

2.4.3 R&D and Productivity Growth in Japanese manufacturing firms

Kwon and Inui (2003) conducted a research to examine the relationship between the R&D and the productivity improvement in Japanese manufacturing firms In this research, they estimated a Cobb-Douglas production function with three inputs: labor, physical capital and knowledge capital for more than 3,000 Japanese firms for the period 1995-1998

The data used in this research is drawn from the Basic Survey of Business Structure and Activities conducted by Japanese Ministry of Economy, Trade and Industry

From this data set, the authors selected 3,830 firms in the manufacturing sector which had positive R&D expenditures from 1995 to 1998 Those firms all have no less than

50 employees and 30 millions yen of capital and are grouped into 22 manufacturing industries based on their main business activities

On investigating the contribution of the R&D to the productivity growth of Japanese manufacturing firms, Kwon and Inui (2003) used two approaches: Production Function Approach and The Rate of Return to R&D Approach In the production function framework, the disadvantage is that possible bias is allowed due to simultaneous output and input decisions, and the advantage is that the assumptions of competitive factor markets, cost minimization, and constant returns to scale are avoided In this approach, the relationship between R&D and productivity growth is represented in the regression function using first-differences as follows:

Where: Y denotes the value added, K as the physical capital stock, L as the labor input, and R as the knowledge capital stock A-t is the time-specific variable and the rate of disembodied technical change The subscripts i and t denote the firm and the year respectively Here, 1 - a > f3 is assumed to maintain a positive marginal product of labor r is a scale parameter r implies increasing returns to scale if it has a positive value, and decreasing returns to scale if it has a negative value

In the second approach, K won and Inui estimated the contribution of R&D to productivity by estimating the rate of return to the R&D rather than the elasticity of value added with respect to R&D The advantage of this approach is that it can avoid the measurement problem of the R&D capital stock The relation between the growth of labor productivity and the level of R&D intensity can be written as the following equation: where E is the R&D expenditures of firm i in period t

In both approaches, the study found a positive and significant effect of R&D expenditure on productivity growth and this effect is different by the firms' sizes and characteristics of technology The R&D elasticities are higher for the large sized and high-tech firms than they are for other types of firms Besides R&D capital, the physical capital stock also significantly affects labor productivity growth Moreover, an industry effect is found not important in explaining productivity differences among firms

2.4.4 The effect of R&D Capital on Danish Firm Productivity

Unlike the two above studies using time series data on analysis, this paper analyses the importance of R&D for Danish private firm productivity on the basis of cross- section data It was conducted by Graversen and Mark (2005) This report aims to identify the return to R&D capital rate and other related factors that have influence on firm productivity growth The data used in this research is drawn from the official Danish R&D Statistics 200 1 conducted by the Danish Centre for Studies in Research and Research Policy The sample contains more than 2200 firms with positive R&D among 18.381 firms in 2001 and it represents broadly all Danish private sector firms with more than 9 employees The analysis is based on a logarithmic version of the Cobb-Douglas production function where productivity is estimated as a function of independent variables as follows:

Productivity= f (R&D Capital, Assets, Labour, Business Sector, Size)

OVERVIEW OF R&D AND FIRM PERFORMANCE IN

INTRODUCTION

The purpose of this chapter is to present the overview of R&D activities in Vietnam in general and R&D activities of firms in particular so that the research is analysed based on a reality background This chapter consists of three main parts First, part one talks about R&D activities in Vietnam and compared with other countries Structure of the R&D system in Vietnam is stated in the next part The final part will discuss how R&D institutions are linked with the productive sector.

R&D ACTIVITIES IN VIETNAM

As mentioned in the rationale of the research, in comparison with OECD countries and neighbors, Vietnam spent a rather little amount on R&D activities In 1996, Vietnam spent approximately 0.3% of its GDP on R&D (figure 3.1) and this number increased to about 0.5% in 2003 But most of such expenditure was financed by the Government, namely 80% in 2002 This is different from OECD countries, where most of R&D expenditure was financed by companies, say about 70% in 2002 (figure 3.2)

Figure 3.1: Percentage ofGDP spent on R&D in 1996

Figure 3.2: Expenditure on R&D by Government and Business sector in 2002

Figure 3.3: Sector-wise R&D Expenditure in Vietnam in 2002

• Direct Government • Business enterprises o Funds from abroad

0 Higher Education • Private non-profit C others

According to Bezanson et al., (2000), NISTPASS 3 provided statistic nwnbers which show that total R&D spending per full-time researcher has been declining sharply since reforms under "doi moi" was introduced For instance, the per labor R&D expenditure decreased from US$687 in 1987 to US$289 in 1990 Even though these estimates may encounter some error, such figures were much lower than that of other East Asian countries, namely US$135,000 in Japan or more than US$50,000 in Korea and Singapore NISTP ASS also said that in the early of 1990s, the annual expenditure on research facilities per full-time researcher was around US$50 Moreover, the number of Vietnamese researchers having opportunities to do with experimental equipment, which is in the same quality with those of most East Asian countries, is approximately 10% Regarding R&D personnel per thousand of total employees, according to Nguyen (n.d.), that number of Vietnam was 0.59 in 2002, higher than Thailand or India, but still much lower than other East Asian countries such as Japan, Korea, Singapore and China

3 NISTPASS is National Institute for Science and Technology Policy & Strategy Studies

Figure 3.4: R&D Personnel per Thousand of Total employees in 2002

STRUCTURE OF THE R&D SYSTEM IN VIETNAM

Bezanson et a!., (2000) stated that the structure of the R&D system in Vietnam consists of three main components The first one is laboratories and other R&D institutes of line ministries or government agencies There are approximately 180 R&D units of this sort in many provinces of Vietnam Except for some state-owned large corporations such as Petro Vietnam, which runs its own labs, Vietnamese industrial enterprises seldom do research & development work by themselves and have little experience on it The second component is universities and colleges

However, the number of university faculties which have sufficient resources for performing R&D activities is rather tiny Most Vietnamese universities and colleges lack of necessary personnel, equipment, libraries and other resources to do the task

And the last one is national research centers (or academies) that are under control of the Government Office Of which, the National Center for Natural Science and Technology (now is Vietnamese Academy of Science and Technology) is the most significant

It is expected that the three main above components have close connections with each other Each one is assigned different function Research institutes of individual ministries are assigned to do applied research and experimental development 4 ; whereas, universities and colleges are the main providers of R&D human resources

Vietnamese Academy of Science and Technology is mainly responsible for performing advanced basic research

However, according to Nguyen and Tran (n.d.), the quality of research infrastructure of Vietnam is lower than international standards The tendency of researches carried out is theoretical, supply-driven and does not meet the manufacturing sector's demands Most of R&D activities in Vietnam are performed in research institutes of ministries and national research centers, instead of universities Moreover, most R&D, which is publicly funded, is carried out in research institutes of the government In Vietnam, there is a small fraction of R&D activities financed by the state budget In general, "the national R&D system is organized, financed and managed in such a way that technology transfer is difficult and expensive" (Bezanson et al., 2000 cited in Nguyen and Tran, n.d.)

Table 3.1: Science & Technology Organizations in Vietnam by 31 Dec 2003

Administration Num % Num % Num % Num %

Source: MOST 5 (2004), cited in Nguyen and Tran (n.d.)

4 The definition of basic research, applied research and experimental development are discussed in the chapter

5 MOST: Ministry of Science and Technology

3.4 LINKAGE BETWEEN THE PRODUCTIVE SECTOR AND R&D INSTITUTIONS

With reference to the relationship between productive sector and R&D institutions, the linkage is so weak SME enterprises often rely on external sources for their R&D, and depend little on R&D institutions and universities Large corporations also do the same as SMEs With regard to foreign sector, most of their R&D activities are performed in home countries of parent firms In general, there are not many chances for Vietnamese R&D institutions and universities to help firms significantly Research institutions in general and universities in particular still face many difficulties in linking with enterprises In many surveys, it is confirmed that firms have demand for technology and training services which are supplied by R&D institutes and universities Nevertheless, this demand has been hardly satisfied R&D institutions and universities on a whole have supported firms at a level which is below desirable (Tran, n.d )

In order to understand more about this issue, it is necessary to know the nature of innovation in Vietnam R&D is one part of innovation 6 In the study of Nguyen and Tran (n.d), they said that Vietnam is a transitional and developing country, therefore, its innovation environment differs from that in developed countries External factors have much influence on innovations of Vietnamese firms Vietnamese innovation environment comprises some main features as follows:

Most Vietnamese firms' competition is mainly based on the availability of natural resources and access to cheap labor There are few enterprises whose competition are based on the background of new technology or differentiated products

Innovations of those firms who make components or operate as subcontractors for foreign companies are under determination of foreign customers

Innovation system is weak at both ãnational and local levels There are limited public resources for R&D and supports for innovation

6 Definition ofOECD, 1994 This was discussed in chapter 2

Markets for technical and innovation services are under development; and so on

Nguyen and Tran (n.d.) also said in their report that "innovations in Vietnam are either incremental or new to the firms" Incremental innovations occur when enterprises attempt to deal with certain technical problems originating from the run of imported production line or when they attempt to make new products by the existing machineries The circumstance when firms purchase a whole or parts of a production line to make new products is called "new to firms" innovations

Thus, it is not surprised that research institutions do not match with production sector in supporting innovation Nguyen and Tran (n.d.) explained that because the nature of Vietnamese firms' innovations is problem-solving, not science-based, academic institutes are expected by firms to help them deal with their technical problems

Nevertheless, academic institutes' organization and priorities do not enable them to do as firms expected effectively Moreover, there is a relatively big difference between what academic institutions are wanted to do by firms and what they think that they should do for firms

Table 3.2: Ranking of most wanted services (for firms) and most capable activities (for academic institutions) of enterprises

Type of services Firms' Rank Academic Insts'rank

Installation of new machines and equipment 2 10

Maintenance and fixing production machines 3 9

Analyzing, testing product/material sample 1 4

Modify product design or material specifications 8 7

Manufacture production machines or components 10 6

Advice in buying production machines 9 3

Source: Innovation Survey 2002 & PROs Survey in 2000 ofNISTPASS, cited in Nguyen and Tran (n.d)

RESEARCH METHODOLOGY

INTRODUCTION

The main objective of this chapter is to design an econometric model, which allows to observe the determinants of productivity growth in general and to measure the impact of R&D activities on productivity growth in specific This chapter includes the following sections Firstly, an econometric model is specified based on relevant theories and empirical studies This section also provides the definitions and measurements relevant to the variables used in the regression model Secondly, given the analytical framework, methods and major problems of data collection are considered The final section is a summary of the research methodology.

MODEL SPECIFICATION

Based on the theories and the empirical studies mentioned above, the relationship between R&D and productivity of Vietnamese manufacturing finns is represented by a model which is a logarithmic version of the Cobb-Douglas production function The production function includes the standard factors such as capital and labour as well as the additional factor, knowledge capital The objective is to attribute the rate of increase in productivity to increases in its inputs Productivity is measured in terms of labour productivity because it relates to the most important factor of production _labour and is relatively easy to measure The assumption of constant returns to scale is also explicitly tested

The Cobb-Douglas specification can be written as:

Dividing both sides by the labor inputs, taking the logs of both sides and adding some interaction variables, we have the regression equation as follows:

Log(-)= a 0 + a 1 1og(-)+ a 2 1og(-)+ a 3 logL + a 4 SIZE + a 5 0WNS + u (4.2)

K denotes the stock of physical capital

Log( Y) denotes natural log of labor productivity based on output

L log(K) denotes natural log of physical capital per labor

L log(R) denotes natural log of R&D expenditure per labor

L log L denotes natural log of people employed SIZE denotes firm size, dummy variable

OWNS is types of ownership, dummy variable u is the error term a3 is a scale parameter If this coefficient is statistically significant and different to zero, we can reject the assumption of constant returns to scale with the respect to three inputs Increasing returns to scale is implied if a 3 has a positive value, and vice versa, decreasing returns to scale is implied if a 3 has a negative value

In this study, due to the limitation of the data material, the cross sectional data is employed to test the relationship between R&D expenditure and productivity

Therefore, important variables in the multiple regression model should be transformed into log terms as well as ratio terms in order to reduce heteroskedasticity and increase the validity of the usual testing procedure Another reason for using cross sectional data is that the R&D elasticity is significant and larger in the cross sectional dimension, whereas it is smaller and usually statistically insignificant in the time series dimension (Matteucci and Sterlacchini, 2004 )

In most of the empirical studies, output is measured by value-added, namely by the sum of operating profit, depreciation allowances, employee compensations, taxes and levies, and rent (Kwon and Inui, 2003) However, due to the limitation of the data set, the calculation of value-added is impossible for this study Therefore, output of each firm (Y) is measured by total sales or revenues instead of value-added Labor (L) is measured by the number of employees because there is no available information on the labor working hours of firms In order to correct for double counting, the R&D employees were subtracted from the total number of employees since R&D manpower is evaluated as R&D expenditure Physical capital (K) is total fixed gross assets; however, fixed gross assets in firms' financial statements are measured by nominal value (book value) According to Cuneo and Mairesse (1983), the physical capital used for R&D expenditure should also be subtracted from the total physical capital stock for correcting double counting However, due to limitation of the data, it is impossible to do the task

With reference to the measurement of R&D capital, since R&D expenditure is available in the data set and it is too complicated to calculate R&D capital according to its definition, the available R&D expenditure is used As mentioned in the literature review, R&D capital is often calculated from R&D spending in the past and R&D spending is accumulated, deflated and depreciated Griliches (1979) stated that R&D capital is considered as a measurement of the current state of technical knowledge and it is determined partly by current and past R&D expenditure (cited in Wang and Tsai,

2003) This means that current R&D capital reflects R&D expenditure both at that time and in the past Ideally, the lag structure that links past R&D expenditure with current increase in technical knowledge should be estimated from the data (Wang and Tsai, 2003) However, in this research, R&D expenditure is available in only one

- - - , - - - - - - - year, therefore, it is impossible to measure R&D in such way This is one of the limitations of this research, which will be discussed in the end

With reference to dummy variables, firm size (SIZE) is categorized by number of employees into three kinds of size: large scale, medium scale and small scale

Moreover, firms are also categorized into three kinds of ownership (OWNS) such as state-owned, foreign-owned and others

KIL is an important input of production that supports productivity growth and its natural log is found to be positively significant in most empirical studies As a result, the sign for log(KIL) is expected to be positive in the productivity function R&D or technical advances allow firms to improve their productivity and this has been proved true by many studies Those firms spending more for R&D activities can have higher productivity Therefore, log(RJL) is expected to be positive Regarding log(L ), it is expected to be positive because the more labor is employed, the more productivity the firm has However, if the number of labor of firm reaches the point where occur the law of diminishing returns, the sign will be inverse Holding other variables constant, it is assumed that large scale firms have higher productivity than other firms and have the same sign with log(L) Similarly, foreign-owned firms are also expected to have higher productivity than state-owned firms and others

In conclusion, the regression equation can be fully rewritten as follows:

Log(-) = a0 + a 1log(-) + a2log(-) + a3log L + a4LARGESCL + a5MEDIUMSCL

Where: LARGESCL and MEDIUMSCL denote large or higher scale and medium scale firms respectively STATE represents state-owned firms FOREIGN represents foreign-owned firms

; a 1 > 0, a 2 > 0, a 3 > 0, a 4 > 0, a 5 > 0, a 6 < 0, a 7 > 0 The signs of the coefficient in version (4.3) result from the argument stated in version (4.2) All signs of the coefficients are based on the empirical findings from the previous empirical studies

It is worth noting about the disturbance term u In addition to the inputs listed in the model, some unobservable factors, such as managerial capabilities, also have considerable impacts on a firm's productivity (Wang and Tsai, 2003) These factors vary across firms, thus, the variances of u may be heteroskedastic This problem will be considered in the estimations since it could result in biases or inefficient estimates.

DATA TRANSFORMATION

The data used in this paper are drawn from the Vietnam Enterprise Survey (VES) conducted by the General Statistics Office in 2004 The data set covers more than 91,750 enterprises in manufacturing, mining, construction, and commerce sectors in all provinces in Vietnam From this data set, only observations in the manufacturing sector are selected among 450 firms having positive R&D expenditure in 2004 7 • Industries in the data set are classified according to the complete list of NACE (Classification of Economic Activities in the European Community) Therefore, it is quite easy to extract only firms belonging to the manufacturing sector Finally, the sample consists of only 264 observations which are qualified for the regression

However, the data must be transformed into an appropriate form before inputting into the model This section describes how variables in version (4.3) are computed based on some assumptions

4.3.1 Labor productivity based on output (Y/L)

7 It cannot be distinguished between firms that did not report their R&D expenditw-es and those that had no R&D expenditures

This variable is the ratio of total output of a manufacturing firm to its total labor in

2004 Total output of a firm (Y) was measured by turnover of goods & services activities which was available in the questionnaire Total labor input (L) was equal to total labor at the end of the year subtracting total R&D employees at the end of the year The labor productivity variable was taken as a natural log and its original unit was VND million

This variable denotes the physical capital per each employee of a firm with its unit as VND million KIL was also taken as a natural log Physical capital (K) was measured by total fixed assets and long-term investments at the end of the year collected directly from the VES-2004

R/L was equal to the ratio of R&D expenditures of a firm to its total labor input with its unit as VND million and also was taken as a natural log R&D expenditures (R) were collected directly from the VES-2004

In this thesis, firm sizes are categorized into small, medium, and large expressed by three dummy variables: LARGESCL, MEDIUMSCL and SMALLSCL According to Small and Medium Enterprises (SME) definition of the EU, an enterprise is classified as medium-size if it has fewer than 250 employees, an enterprise with less than 50 employees is regarded as small, and an enterprise with less than 10 employees is considered as micro (European Commission, n.d.) Thus, LARGESCL denotes firms with headcounts of equal or over 250 (>%0) while MEDIUMSCL denotes those with headcounts of fewer than 250, but equal or over 50 (50 - 249) and the rest (1 -

49) is regarded as SMALLSCL However, only two dummies, LARGESCL and MEDIUMSCL are included in the regression model

4.3.5 Types of ownership (STATE, FOREIGN)

STATE denotes firms having more than 50% of shares owned by the state FOREIGN denotes firms owned by the foreign including 100% foreign invested enterprises and joint-venture enterprises with one partner as the foreign.

RESULT ANALYSIS

INTRODUCTION

The main purpose of this chapter is to analyze findings from the regression model in order to answer the research question that whether there is a positive relationship between R&D expenditure and productivity growth in Vietnamese manufacturing firms This chapter consists of three sections The first section describes main characteristics of firms in the sample by descriptive statistics method In the next section, a regression model which has been specified in the previous chapter is run and tested Then, model results, empirical findings from the model and its statistical validity are analyzed and explained in a reasonable way The last section contains a summary of the result analysis.

FIRMS CHARACTERISTICS

The data set used for sampling in this study contains 91,7 55 observations, among which there is only 450 firms (0.49%) in all business sectors having positive R&D expenditure From such 450 firms, only 264 firms in the manufacturing sector are chosen for the sample In the population aspect, it means 0.49 percent of firms in all sectors and 0.29 percent of firms in the manufacturing sector have positive R&D expenditures Even though it is impossible to distinguish between the firms that did not report their R&D expenditures and those that had virtually no R&D expenditures, and such numbers may be larger, they still indicate a fact that there is a small fraction of firms in Vietnam who care and invest in R&D activities

Figure 5.1 shows that in the selected sample, most of observations, which account for 59.47 percent, are owned by the state; whereas, only 11.74 percent are owned by foreigners or joint-ventures According to the EU's definitions of SME, figure 5.2 presents that large firms (>= 250 employees) make up more than 60% of the sample

The next is the medium (50- 249 employees) and the small (1- 49 employees) with 25.38% and 13.26%, respectively

Figure 5.1: Structure of firms by ownership

Figure 5.2: Structure of firms by size

Source: Author's calculation based on the data ofVES-2004

Based on the industrial classification of NACE, the number of firms in each industry and its percentage in the sample have been calculated and presented in the table 5.1

According to that, there are up to 4 7 firms manufacturing Chemicals and Chemical Products, which account for the largest share of the sample (17.80%) The second largest share of the sample (17.05%) goes to 45 firms which produce Food Products and Beverages Moreover, there are 26 firms operating in the Other Non-Metallic Mineral Products manufacturing, which make up 9.85%, third after Food Products and Beverages The Machinery & Equipment manufacturing and the Electrical

Machinery and Apparatus manufacturing also make up rather large shares of the sample with 7.95% and 6.82%, respectively Other industries just account for small fraction in the sample There is one point that the technology of most manufacturing firms in the sample lie somewhere between low and medium level

Table 5.1: Industrial Classification of the Sample

:No~,:: , lndustl)' ã ã NC) or firms Percent (o/0)

4 Wearing Apparel, except fur apparel 9 3.41%

5 Manufacture of luggage, handbags, saddlery, harness and

6 Wood and Products of wood and cork; Articles of straw and

11 Other non-metallic mineral products 26 9.85%

13 Fabricated metal products, except machinery and equipment 10 3.79%

17 Radio, Television and Communications Equipment and

18 Medical, Precision and Optical Instruments, Watches &

19 Motor Vehicles, Trailers & Semi-Trailers 7 2.65%

Source: Author's calculation based on the data ofVES-2004

Table 5.2 presents that state-owned enterprises have larger average number of labour than other types and the full sample The firm having the largest number of labour (8,993) is also a state owned In spite of large scale in labour, state owned enterprises perform less efficiently than foreign enterprises The average turnover of foreign firms is VND381,314 million, higher than VND327,332 million of state owned firms

Moreover, foreign firms are stronger than state owned firms and other types of firms in capital resource and they also pay much more attention to innovation, research and development of technology and science The total fixed assets and long-term investment of the foreign is VND193,764 million on average; whereas, the number of the state owned is VND113,795 million The average R&D expenditure of the foreign (VND7,143 million) is seven times higher than that of the state owned (VND1,076 million)

As categorized by size, large firms obviously have larger mean turnover and capital amount than medium and small firms The difference between mean turnover of large firms (VND391,028 million) and that of the small (VND10,500 million) and the medium (VND87,114 million) is quite big It is about 4 times and 35 times for medium firms and small firms, respectively However, mean R&D expenditure of medium firms is VND1,783 million, a little bit higher than that of large firms, VND1,749 million The firm which has the largest amount of R&D expenditure, namely VND55,210 million is a foreign owned and large-scale one

142,518.0 52,148.0 676.5 300.0 Source: Author's calculation based on the data ofVES-2004

Note: The unit of Y, K and R is VND million

According to figure 5.3 and 5.4, most of the total cost for research and development of science & technology of manufacturing firms in the sample are from their own investment with 82 percent Whereas, only 7 percent and 1 percent of total R&D cost are financed by the state budget and from foreign countries, respectively That means enterprises have to finance for technology improvement through research & development by themselves However, 81 percent of the cost is for developing technology and only 17 percent is for research purpose This can be easy to explain because the nature of Vietnamese firms' innovation is problem-solving, not science- based (Nguyen and Tran, n.d.) Firms' R&D activities seem to mainly focus on such problem-solving works as installing a new production line, applying a new technology or improving the existing technology to produce new products

Figure 5.3: Total cost for research & development of technology by resources

Source: Author's calculation based on the data ofVES-2004

Figure 5.4: Total cost for research & development of technology by purposes

Source: Author's calculation based on the data ofVES-2004

REGRESSION ANALYSIS

In this section, two main issues will be discussed Firstly, the correlation matrix is analyzed to examine whether independent variables are suitable to be entered into the regression model Secondly, the model estimation, its empirical results and the statistical validity of the model are presented

5.3.1 Correlation matrix Table 5.3: Correlation matrix from the variables in the function

FOREIGN 0.38 0.35 0.33 -O.IO ã Source: Author's estimate based on the data ofVES-2004

All Ln denote natural log of variables;

Y _ L denotes labor productivity based on output;

K _ L denotes physical capital per labor;

L denotes the total number of labor;

FOREIGN denotes type of ownership (Foreign firm= I, if not= 0);

STATE denotes type of ownership (State-owned firm= I, if not= 0);

LARGESCL denotes firm size based on the number of labor (Large = I, if not = 0);

MEDIUMSCL denotes firm size based on the number of labor (Medium = 1, if not = 0);

According to the relationship between variables of the productivity model shown in the table 5.3, the three important independent variables (lnK_L, lnL, lnR_L) have i positive correlation with the labor productivity These variables are compulsory in this productivity model Among these independent variables, physical captial per labor have stronger relationship with productivity than others Moreover, as expected, the correlation between foreign firms and productivity is higher than that between state- owned firms and productivity

5.3.2 Model estimation and empirical results

Labor productivity is influenced by physical capital per labor, number of labor, R&D expenditure per labor, size and ownership Therefore, an analysis of these institutional main factors is necessary in determining the influences to labor productivity Table 5.4 shows the empirical results of the productivity model Except STATE, other variables are statistically significant at 1%, 5% and 10%

Firstly, dummy variables are examined to see how productivity growth differs according to ownership and firm size With reference to the ownership, only FOREIGN variable is statistically significant at 10% Its coefficient indicates that when other variables are constant, mean productivity growth rate of foreign firms is 1.3% higher than that of state-owned firms and others This is easy to understand because state-owned enterprises usually operate in a less efficient way than the foreign, which has been discussed in the descriptive statistics of firms above There is no difference in the productivity growth rate between state-owned firms and other firms such as private companies, limited companies, etc In relation to firm size by the number of labor, the regression results suggest that the mean productivity growth rate of both medium-sized (50-249 employees) and large-sized firms (>249 employees) are higher than that of small-sized firms, holding other variables constant Moreover, productivity growth rate of medium-sized firms is a little bit higher than that of large- sized firms

Table 5.4: Coefficients and statistics for the productivity model

Unstandadized Coef Standadized Level of t Prob

Ln of Physical capital per labor (LnK_L) 0.349 0.005 0.49 72.326 0.000 1%

Ln ofR&D expenditure per labor (LnR_L) 0.102 0.009 0.16 11.859 0.000 1%

Source: Author's estimate based on the data ofVES-2004

Standadized coefis the correlation coefbetween each independent variable and dependent variable

The physical capital per labor (K _ L ), R&D expenditure per labor (R _ L) and total labor (L) are factors which have direct impacts on productivity based on the arguments of the Cobb-Douglas production function and the R&D capital model

These factors were taken as the natural log to represent proxy variables in the model

Each coefficient of these three variables is the elasticity of labor productivity The regression results suggest that all those variables have positive effects on productivity, which is similar to the expectation Among these three variables, physical capital per labor has the strongest impact on productivity The elasticities of productivity with respect to physical capital per labor and total labor were around 0.3 5 and 0.15, respectively These numbers can be said in another way that holding other inputs constant, a 1% increase in the physical capital per labor and total labor lead to a 0.35% and 0.15% increase in productivity, respectively

5.3.2.2 Analysis of the elasticity of productivity with respect to R&D Being lower than the elasticities of productivity with respect to the two other variables, the one with respect to R&D expenditure per labor was about 0.1 This

- - - - - - number means that a 1% increase in R&D expenditure per labor leads to only 0.1% increase in productivity This number is acceptable because Griliches (1995) demonstrated in a cross-sectional study that the output elasticity of R&D capital was around 0.09-0.14 (cited in Wang and Tsai, 2003) Thus, physical capital and labor have more effects on productivity than R&D expenditure This can be explained that Vietnamese enterprises do not care much about investing for R&D while labor is a cheap and easily-accessed input

Even though the result that R&D expenditure has positive effect on productivity in this case, there are some points needed to be discussed Wang and Tsai (2003) said that the conclusion that R&D investment has a positively significant impact on productivity in most studies may be overoptimistic due to the problems of 'file- drawer' and measurement of R&D capital As mentioned before, due to limitation of the data, this study fails to measure R&D capital in a way other studies often do It neglected the accumulation of R&D expenditure in the past, its lag, deflation and obsolescence However, the regression result is a surprise to the author It may be explained that because most R&D activities of Vietnamese enterprises are problem- solving, not science-based; and most of R&D expenditure are for developing technology, not for research (Figure 5.4) Such technology-developing activities have direct effects on technology and therefore on productivity more quickly than research activities do

5.3.2.3 Statistical validity of the model

In this model, the hypothesis of constant return to scale with respect to three inputs (K, L, R) is tested The coefficient of natural log of total labor or a 3 (see model4.2) is a scale parameter This coefficient is statistically significant and around 0.15, which implies that the assumption of constant returns to scale is rejected at the significant level of 1% and increasing returns to scale is accepted

There are two important tests in this model: multicollinearity and heteroscedasticity

These tests are to make sure the statistic validity of the model It is unnecessary to test multicollinearity problem for the model because most of the correlation coefficients between regressors did not exceed 0.5, except some correlation coefficients relating to dummy variables (Table 5.3) Moreover, independent variables included in the model were statistically significant at 1% and R square of the model were lower than 0.8 (Table 5.4 and Appendix 2) 8 Regarding heteroscedasticity, the model was handled to be free of it Detailed results of White Test are shown in the Appendix 3.

CONCLUSIONS AND RECOMMENDATIONS

CONCLUSION

In this study, the relationship between R&D expenditure and productivity of manufacturing firms has been analyzed based on a sample of 264 manufacturing firms reporting positive R&D The sample was drawn from the data set including 91,755 observations in all sectors of the Vietnam Enterprise Survey in 2004 In the sample, state-owned firms account for 59.47% while foreign firms account for only 11.74%

Large scale firms (>249 employees) make up more than 60% of the sample, the medium (50 - 249 employees) and small (1- 49 employees) make up 25.38% and 13.26%, respectively With reference to the industrial classification, firms manufacturing Chemicals and Chemical Products account for the largest share of the sample (17.08%) The second largest share belongs to firms operating in the Food Products and Beverage manufacturing (17.05%)

The statistic summary indicates that state-owned enterprises seem to perform in a le~s efficient way than foreign ones on average Furthermore, foreign firms seem to be stronger in capital resource and invest more in innovation, R&D activities than state- owned firms and the others Regarding the structure of total cost for research and development of science & technology of manufacturing firms in the sample, 82% of the cost is financed by firms themselves Nevertheless, 81% of such cost is used for developing technology and just 17% is used for research

In relation to the analytical framework of the research, the regression equation was estimated mainly on the basis of the Cobb-Douglas production function and the R&D capital model It was formulated in a logarithmic form and had statistical validity The regression model can be written as follows:

Log(y) = 2.24 + 0.351og(K) + O.IOlog(R) + 0.15log L + 0.26LARGESCL +

The most important finding of the research is that R&D investment was a significant determinant of growth of firm productivity Holding other variables constant, I% increase in R&D expenditure per labor leads to about 0.1% of labor productivity growth Even though R&D capital was measured in a much si"mpler way than most related studies usually do by using available R&D expenditure in the survey, the result is so surprised It is due to the statistical significance and positive sign of R&D elasticity coefficient The reasons may be that most R&D activities of Vietnamese enterprises are problem-solving, not science-based and most of R&D expenditure are for developing technology, not for research

In addition to R&D capital, other independent variables are also statistically significant The elasticities of productivity with respect to physical capital per labor and total labor were around 0.35 and 0.15, respectively Physical capital per labor and labor have more effects on productivity growth than R&D expenditure per labor

Moreover, ownership and firm size also have impacts on productivity growth rate If other variables are held constant, it is found that the productivity growth rate in foreign firms is higher than in state-owned and other firms There is no difference between state-owned firms and others due to the insignificance of STATE variable

Similar to foreign firms, productivity growth rate of medium-sized firms (50-249 labors) is a little bit higher than that of large-sized firms (>249 labors) and much higher than small-sized firms

In conclusion, besides usual production inputs such as physical capital and labor, R&D activities also plays an important role in productivity growth of manufacturing firms in Vietnamese case Hence, the first research question as mentioned in the first chapter is answered Regarding the second research question, manufacturing firms should pay more attention on investment in R&D activities in order to increase their productivity The section below will discuss some policy recommendations to support firms.

POLICY RECOMMENDATIONS

In order to stimulate R&D activities in the industry sector, the Korean Government used a variety of methods and policies which were flexible to different periods and national development strategies In 1960s and 1970s, polices mainly focused on tax motivation and preferential treatment to R&D activities However, these policies' results were below expected level because firms in the industry sector lacked of clear demand for investment in R&D and they thought that it was easy to access prevailing technology from many sources In 1980s, there were other motivation methods such as: decreasing tax rate applied for importing R&D equipment; spending for R&D and development of R&D human resources were considered as activities deducted from tax; or exempting fixed assets relating to R&D from taxes, and so on

Moreover, the Korean Government launched some indirect programs stimulating R&D activities such as International Standard Korean Products Program; or the Government built a list of 21 products relating to 59 manufacturers in the industry sector to provide supports Besides tax motivation methods, the Government also had financial supporting policies to encourage enterprises in investment for R&D activities Small enterprises, which were unable to establish their own R&D centers, were encouraged to cooperate with other firms Thanks to such solutions, the number ofR&D institutes and associations increased sharply, especially in 1980s- 1990s

In September 1999, Korean Industrial Property Office launched a campaign to support SMEs This program aimed at encouraging all SMEs to invent new technologies and use them as their key business assets In order to achieve the target, the program was conducted in such steps as strengthening the community's awareness of intellectual

; property; connecting R&D activities with intellectual property; improving patent- issuing procedure; and supporting the usage, transaction of patent technologies

Furthermore, in order to support financial resources for the commercialization of new technologies, the Government stimulated the development of private venture capital and initiated by establishing government venture funds The Government also enhanced capital market for newly-established firms, science and technology firms; and developed a secondary stock market for them

Majority of Vietnamese firms now are small or medium in scale, weak in financial resources; therefore, investment for R&D activities is limited Besides, the domestic market for technology services has not developed, firms do not have information about researches, inventories conducted in the country Therefore, many firms just take into consideration the acquisition and development of new equipment, and ignore research or technology transfer SME enterprises often rely on external sources for their R&D because local R&D institutions have not well connected with enterprises

For many reasons, R&D activities have not been invested at a proper level

However, in the context of international integration today, enterprises have no choice to survive and develop by investing for R&D activities and technology innovation as well These activities may help firms improve production capabilities, which lead to productivity growth, production costs saving, lower price of products and higher competition ability Even though these activities may take times, require a large amount of capital, and suffer high risks, firms must be determined to conduct these activities because it is a long-term investment In addition to firms' own attempts, supports from the authority are necessary to stimulate their R&D investment Thus, in this research, some policy recommendations are drawn based on the experience of Korea and the actual context of Vietnam to encourage firms to invest in R&D activities as follows:

Equitizing R&D institutions so that they can be more active, creative and responsible in their activities to meet demand of the productive sector As mentioned before, the research infrastructure of Vietnam is lower than international standards, the link between the productive sector and research institutions is weak Local enterprises depend on external sources for R&D and foreign ones rely on their parent companies in home countries In general, the support of R&D institutions and universities to firms is under desirable level

Equitization may enable them to operate in a more market-oriented and efficient way Through this, R&D institutions may have better contribution in supporting enterprises

Focus on tax motivation and preferential treatment to R&D activities such as decreasing tax rate applied for importing R&D equipment; spending for R&D and development of R&D human resources are considered as activities deducted from tax; exempting fixed assets relating to R&D or Technology Development Fund from taxes, etc

Establishing R&D venture funds of the government to support financial resources for firms and stimulate the development of private venture capital

These funds must be managed and supervised strictly to make sure that the capital is used in right places, at right time and for right purposes

Stimulating firms to establish their own R&D departments or cooperate with other firms This policy should be specified by programs or campaigns, which are conducted in many years and at the national level.

LIMITATIONS OF THE RESEARCH

The research does of course have some limitations First of all, most studies regarding the contribution of R&D to productivity growth suffer the double counting of R&D labor and physical capital, which are counted once in measuring labor and physical capital and once agam m measurmg R&D capital (Cuneo and Mairesse, 1983)

However, the research subtracted only R&D labor from the total labor and failed to correct the double counting of physical capital In reality, the task is impossible because the limitation of the data

Secondly, due to the availability of data in only one year 2004, the research failed to measure R&D in a way it should have been done R&D capital was measured by available R&D expenditure in the data The research ignored the accumulation of past R&D spending, its lag, deflation and depreciation Moreover, the relationship between R&D expenditure and productivity growth was examined for only 1 year, therefore, the result of R&D elasticity is still skeptical This fact is the suggestion for another research in the future, which use panel data or time-series data, instead of cross- sectional data Measurement of R&D expenditure in many years may bring a more reliable result than the recent study does

Bezanson et al (2000) A Science Technology and Industry Strategy for Vietnam [online] Available: http://www.unido.org/fileadminlimport/20398_kbfin.pdf

Congressional Budget Office (2005) R&D and Productivity Growth [online]

Available: http://www.cbo.gov/ftpdocs/64xx/doc6482/06-17-R-D.pdf

Cuneo, P and Mairesse, J (1983), Productivity and R&D at the firm level in French manufacturing [online] Available: http://www.nber.org/papers/w1068.pdf

Eris, E D and Saatcioglu, 0 Y (2006) A System Look For Technological Innovation:

Firm Based Perspective [online] Available: http://www iseing.org/ emcis/EMCIS2006/Proceedings/Contributions/C52/CRC/ eris%

Graversen, E K and Mark, M (2005) The Effect ofR&D Capital on Firm Productivity [online] Available: http://www.cfa.au.dk/Publikationer/Working_papers/WP2005 _3.pdf

Griliches, Z (2000) R&D, Education and Productivity, London: Harvard University Press

Gujarati, D N (2003) Basic Econometrics, Fourth Edition, International: McGraw-

Kwon, H U and Inui, T (2003) R&D and Productivity Growth in Japanese Manufacturing Firms [online] Available: http://www.esri.go.jp/jp/archive/e _ dis/e _ dis050/e _ dis044a.pdf

Mairesse, J and Sassenou, M (1991) 'R&D and Productivity: A Survey of Econometric Studies at the Firm Level', NBER Working Paper No 3666 Matteucci, N and Sterlacchini, A (2004) ICT, R&D and Productivity Growth:

Evidence from Italian Manufacturing Firms [online] Available: http://www.isae.it/Matteucci Sterlacchini ICT.pdf

Nguyen, M.Q (n.d.), The Role of Technology Upgrading to Enhance Growth and Competition: The Case of Vietnam Available: http:/ /info worldbank.org/etools/ docs/library/23 8724/8%20AIF-Vietnam-Quan.pdf

Nguyen, V.H., and Tran, N.C (n.d.) The Role of Academic Institutions in Economic Development: The Case ofVietnam [online] Availabe: http://www fpi.lu.se/ _ media/en/researchluniversidad06-vietnam pdf

OECD (1994) Main Definitions and Conventions for The Measurement of Research and Experimental Development (R&D) [online] Available: http://www 1.oecd.org/ dsti/ sti/stat -ana/prod/ e 94-84 pdf

OECD (2001) Measurement of Aggregate and Industry-Level Productivity Growth [online] Available: http://www.oecd.org/dataoecd/59/29/2352458.pdf

OSTP (1997) Chapter 2: The Role of R&D and The Changing R&D Paradigm [online] Available: http:/ /belfercenter.ksg.harvard.edu/files/pcast97 ch2.pdf

Pindyck, R S and Rubinfeld, D L (1992) Microeconomics, Second Edition, Singapore: Macmillan

Rogers, M (1998) The Definition and Measurement of Innovation [online]

Available: http://www.melboumeinstitute.com/wp/wp 1998n1 O.pdf

Tran, N.C (n.d.) Vietnam's Innovation System: Toward a Product Innovation Ecosystem [online] Available: http://crds.jst.go.jp/GIES/archive/GIES2006/participants/abstract/33 _ ca-ngoc- tran.pdf

US Census Bureau 'NAICS Sector: 31-33 Manufacturing' [online] Available: http://www.census.gov/epcd/ec97 /def/31-33.HTM f

Wang, J C and Tsai, K H (2003) Productivity Growth And R&D Expenditure In Taiwan's Manufacturing Firms [online] Available: http://www nber.org/papers/w9724 http://www.ic.gc.ca/canadian industry statistics/cis.nsf/IDE/cis31-33defe.html

European Commission (n.d.) The new SME definition: User guide and model declaration [online] Available: http://ec.europa.eu/enterprise/enterprise_policy/sme_definition/sme_user_guide.pdf

A System Model for Technological Innovation

lmpl•m•ntation UNCERTAJNITY CONTMUOUS NATURE

De