The impacts of export and foreign direct investment on total factor productivity evidence from cross country analysis

We study the role of export and foreign direct investment on total factor productivity growth in a sample of both developed and developing countries from 1996 to 2009..  After that, I u

VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

THE IMPACTS OF EXPORT AND FOREIGN DIRECT INVESTMENT ON TOTAL FACTOR PRODUCTIVITY: EVIDENCE FROM CROSS

COUNTRY ANALYSIS

BY

QUAN MINH QUOC BINH

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

HO CHI MINH CITY, MARCH 2012

VIETNAM - NETHERLANDS PROGRAMME FOR M.A IN DEVELOPMENT ECONOMICS

THE IMPACTS OF EXPORT AND FOREIGN DIRECT INVESTMENT ON TOTAL FACTOR PRODUCTIVITY: EVIDENCE FROM CROSS

COUNTRY ANALYSIS

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

MASTER OF ARTS IN DEVELOPMENT ECONOMICS

“I certify that the substance of this thesis has not already been submitted for any degree and have not been currently submitted for any other degree

I certify that to the best of my knowledge and help received in preparing this thesis and all sources used have been acknowledged in this thesis.”

QUAN MINH QUOC BINH

The process of writing a thesis is a collaborative experience involving the support and helps from many people I want to express my gratitude to those who give me the tremendous support to complete this thesis

I am deeply indebted to my parents for their invaluable support and constant encouragement From my early childhood, my parents always teach me valuable lessons on the importance of learning The boundless love of my parents have accompanied with me as

I continue my long journey on the pathway of intellectual acquisition

Associate Professor Doctor Pham Hoang Van, Lecturer at Department of Economics, Baylor University (USA), Fulbright Scholar, is a superb supervisor He has encouraged me

to pursue this topic from the initial ideas to the final completion His wholehearted guidance, incredible patience, and useful discussions have enabled me to develop a deeply understanding on my thesis

Equally, I wish to express my heartfelt gratitude to Associate Professor Doctor Nguyen Trong Hoai, Lecturer at Department of Economic Development, University of Economics (HCMC), my second supervisor, for his valuable suggestions during the time I write this thesis His wide knowledge, excellent advice and logical way of thinking have provided me a good basis in present this thesis

I also take this chance to convey my warm and sincere thanks to Professor Arjun Singh Bedi who have a remarkable influence on my love of academic research His constructive comments have great value in my study

During my time at VNP, I receive a great encouragement and help from many friends, and I am grateful to Ms Ngo Hoang Thao Trang, Ms Chau Ngoc Thao Nguyen, my close friends, for motivating me to overcome all difficulties in my life In addition, my special thanks should send to Mr Nguyen Dinh Quy (Programme Librarian) and Ms Tang Thi Xuan Hong (Programme Secretary) for their help and sharing of their resources to complete this thesis successfully and in time

Finally, I take a pride in myself for working very hard to finish this thesis I realize that after each success stories, there is a process of lots of hard works, difficulties, obstacles and overcoming Even in the hardest time when I write this thesis, I always believe that great efforts will eventually to be paid off Thank for the great time to learn and to grow up

We study the role of export and foreign direct investment on total factor productivity growth in a sample of both developed and developing countries from 1996 to 2009 We start

by employing growth accounting exercise to estimate TFP growth of 103 countries and find out it determinants We also provide a global picture about TFPG’s performance of countries in the world By pointing out the limitations of previous researches that fail to take into account the potential of endogeneity between FDI and TFPG as well as export and TFPG, this thesis contributes to the current literature by developing two new instrumental variables for export and FDI to overcome endogeneity problem Two new instrumental variable have proved their highly effectiveness in removing endogeneity bias We carefully check the robustness of our findings by employing panel data techniques to reestimate the model that used in cross section exercise We also employ one lagged value of FDI and export to control for problem of reversed causality running from TFPG to export and FDI

In empirical analysis, my research finds a robust and positive statistically significant association between export and TFP growth The finding emphasizes the indispensable role

of export in enhancing TFP growth Interestingly, my thesis asserts the existence of negative linkage between FDI and TFPG This finding is sharply contrasts with conventional wisdom

of many people who think that FDI inflows will benefit for TFP growth

From policy perspective, we recommend that a sound macroeconomic stability, moderate government expenditure together with policies to foster export-oriented industries are always needed for a better performance in TFP growth We also suggest that government should invest more in education to increase in quality and skill of human capital Government also needs to spend more money in training programs for the workforce to learn how to apply advanced technology into production

FDI Foreign Direct Investment

FEM Fixed Effect Model

OLS Ordinary Least Square

R&D Research and Development

TFP Total Factor Productivity

TFPG Total Factor Productivity Growth

2SLS Two Stage Least Square

CHAPTER I: INTRODUCTION 1

1.1 Problem Statement 1

1.2 Research Objectives 2

1.3 Research Questions 2

1.4 Research Methodology 2

1.5 Organization of The Study 3

CHAPTER II: LITERATURE REVIEW 4

2.1 Total Factor Productivity 4

2.2 Theoretical Background 5

2.2.1 Exogenous Growth Theory 5

2.2.2 Endogenous Growth Theory 6

2.2.2.1 Learning by Doing Model 6

2.2.2.2 R&D Model 7

2.3 Determinants of TFP & TFPG 9

2.4 Empirical Studies 17

2.5 Conceptual Framework 24

CHAPTER III: RESEARCH METHODOLOGY 26

3.1 Methods in TFP Level and TFP Growth Measure 26

3.1.1 The Regression Method 26

3.1.2 The Growth Accounting Method 27

3.2 Data Source 28

3.2.1 Data for Dependent Variable TFPG 29

3.3.2 Definition of Variables used in Regression 35

3.3.3 Model for Panel Data 40

CHAPTER IV: EMPIRICAL ANALYSIS 42

4.1 Overview of TFPG Performance in The World 42

4.2 Empirical Result 47

4.2.1 Empirical Result from Cross Section Data 47

4.2.2 Empirical Result from Panel Data 51

CHAPTER V: CONCLUSION AND RECOMMENDATIONS 55

5.1 Conclusion 55

5.2 Policy Recommendation 56

5.3 Limitations and further research 57

REFERENCES 59

APPENDIX 65

LIST OF TABLES

Table 2.1: Summary of empirical studies relating to the determinants of TFPG 18

Table 3.1: The definition of variables in the model 39

Table 4.1: Cross Country Regression About The Determinants of TFPG 48

Table 4.2: Panel Data Regression For The Determinants of TFPG 53

Table A1: List of 103 countries in my sample 65

Table A2: TFP growth (%) of 103 countries by regions from 1996 to 2009 67

Table A3: Ranking of average TFPG rates (%) of 103 countries 2005-2009 71

Table A4: Descriptive Statistics of Variables 73

LIST OF FIGURES Figure 2.1: Conceptual framework of determinants of TFPG 25

Figure 4.1: Trend of Total factor productivity growth 42

LIST OF APPENDICES Appendix 1: List of 103 countries in my sample 65

Appendix 2: TFP growth (%) of 103 countries by regions 67

Appendix 3: Ranking of average TFP growth rates 71

Appendix 4: Descriptive Statistics 73

Appendix 5: Regression Result From Cross Section Data 74

Appendix 6: Regression Result From Panel Data 79

Appendix 7: Statistical Tests 83

Chapter I INTRODUCTION

1.1 Problem Statement

It is widely believed that productivity or efficiency of an economy is the most important determinant of income in the long run Solow (1956) explained that economic growth without technological progress (one source of productivity gains) can not be sustained and would be stopped in the long run Parente and Prescott (2004) and Hall and Jones (1996) show that productivity differences explain the large part of income differences across countries Because of the importance of productivity to income, many scholars and researchers have studied possible factors that can affect productivity

In recent years, cross border investment and trade activities have increased remarkably despite the severe impacts of global financial crisis In 2010, global foreign direct investment (FDI) inflows increase $1.24 trillion, and it is expected to rise further towards $1.6 - $2 trillions in 2012 (UNCTAD, 2011) More importantly, FDI and trade are considered as important sources of economic growth, especially for developing countries However, the empirical question whether FDI and trade benefit for productivity growth in different countries at different stage of development is still a question of debated

In this research, I study differences in productivity growth across countries in the world from 1996-2009 and possible factors that affect productivity growth In particular, I look at export and foreign direct investment (FDI) as two channels through which efficiency and productivity are affected If we think of productivity and efficiency as reflecting technology and production knowledge, then exports and FDI are ways that best practices from the outside can be transmitted to the domestic economy While the impacts of export and FDI on economic growth have got plenty attention by many scholars, the research on the impact of these factor on TFP still lags behind Until now, there are few studies of the impact of export and foreign

direct investment on TFP, and those empirical researches that take into account the linkage of export or FDI on TFP fail to deal with endogeneity problem My paper contribute to the current literature by constructing new instrumental variables for export and FDI, and we use a largest data set which includes more countries with additional years, especially taken into account the impact of global financial crisis

on TFPG

1.2 Research Objectives

My main research objectives are listed below:

 To measure productivity growth in 103 countries (including Vietnam) in the period 1996-2009

 After that, I use data of productivity growth at the country level which is calculated above to examine whether exporting and FDI has played any role in increasing productivity in the world

 To suggest policies to speed up total factor productivity growth of countries in the world

1.3 Research Questions

In order to achieve these objectives, my thesis is in an effort to find out answers for the three following research questions:

a How does productivity growth vary across countries?

b Do exports have a positive impact on productivity growth at country level?

c Does FDI has a positive impact on productivity growth at country level?

1.4 Research Methodology

My research intended to use quantitative method by analyzing data of 103 countries from 1996 to 2009 First, I employ cross section data analysis to find out the impacts of FDI and export on TFP I also address the problem of endogeneity which is the central of our empirical analysis Because of the possibility of endogeneity between export, FDI and TFP I will instrument export by the land area variable and FDI by lag FDI, distance from equator (latitude) and land area variables, and I will run two stage least square regression (2SLS) with these

instruments As a more comprehensive ways of robustness check, I also report the result from panel regressions The use of panel data has many advantages such as

we could control for unobserved heterogeneity and to rule out the bias of omitted variable

1.5 Organization of the study

My paper consists of five main chapters After this introduction chapter, the remaining of this thesis is arranged as follow Chapter 2 is the theoretical framework to make sure that my research is built on a truly of scientific knowledge

It widely discuss about definition of TFP and TFPG, present theoretical insights as well as empirical works of previous scholars Chapter 3 is the research methodology In this chapter, I will present methods to estimate TFP and TFPG, and I also set up models to find out TFPG’s determinants Chapter 4 is the empirical analysis The chapter first begin by present a global picture about performance of TFPG of countries in the world Next, the regression results from cross section and panel data is presented Chapter 5 draws conclusion of this research and recommend policies in order to help law makers have better policies for Vietnam

Chapter II LITERATURE REVIEW

I begin this chapter by discussing some definition which involve to TFP and TFPG Next, the theoretical background of TFPG along with its determinants is presented And then it moves to the insights of empirical works in order to exhibit the important role of FDI and export in TFPG

2.1 Total Factor Productivity

Shim and Siegel (1992) defines productivity is “output per unit of input employed” Similarly, Hulten (2009) defines productivity as the ratio of real output

to a unit input As we measure real output per unit of unique input including capital, labor, we have definitions of “factor productivity” (such as capital productivity or labor productivity) When we combine productivity of all factors of inputs, we have definition of total factor productivity (Kopleman, 1986, p.3)

Comin (2008) states his definitions about total factor productivity in “The New Palgrave Dictionary of Economics” as “Total factor productivity is the portion

of output not explained by the amount of inputs used in production” Basing on this definition, TFP represents how efficient and intense inputs are used in the production to generate outputs

Solow (1956) calls the output growth which can not be interpreted by input growth as Total factor productivity growth (TFPG) Specifically, TFP growth is considered as “Solow residual” from the production function For example, TFP growth can be derived by the subtraction of the growth rate of output and the growth rate of input He provides an insightful explanation for the differences in cross-country TFPs which is caused by the differences in technology In turn, differences in technology across country will determine the differences in per capita income of countries

2.2 Theoretical Background

All of the theories related to TFP and TFPG are created by researchers who attempt to understand the growth process In order to have a clearer picture about TFP and TFPG, we should examine the underlying economic growth theories and models However, we only briefly review some milestones in theory that related to TFP and TFPG Almost literature of TFPG is originated from a long debate between neoclassical economists and new growth economists

2.2.1 Exogenous Growth Theory

Exogenous growth theory, which is known as “neoclassical growth theory”, was independently developed by Nobel Prize winner Robert Solow (1956) and prominent economist Trevor Swan (1956) Their model becomes a main branch of economic growth theory during the years of 1950s and 1960s Exogenous growth model tries to explain for economic growth in the long run by exploring factors like capital accumulation, the growth of population as well as productivity

Exogenous growth model begins with a production function as presented below:

Y = A(t)F( K, L ) (2.1) with Y standing for output or income, K is the capital input, L is the labor input A(t) is the technological level or knowledge level, and A(t) is a function of time In this model, A(t) is assumed to grow at exogenous rate Labor force L is also assumed to grow at constant rate

From equation (2.1), we can produce the expression for output per capita

)()(

L

K F t A L

From this equation, we can recognize that output per capita depend upon

capital per capita (

L

K

) and level of technology or knowledge A(t) However, capital per worker is assumed to exhibit a diminishing return That is, when we continue to increase K, the contribution of K to output growth will decrease We can think a

gradually extra equipment become redundant and marginal productivity of capital will decrease Hence, to have a positive output growth per capita in the long run, then technology is a key determinant

Neoclassical economists consider “technology progress” as a main source to interpret for differences in income per capita of countries as long as determine long run growth of economies For instance, Solow (1956) explains that approximately

90 percent of income per capita growth in US is due to exogenous technology progress In a research for the contribution of physical capital, human capital and technology to income difference between 127 countries in the world Hall and Jones (1999) find that technology progress contributes 8.3 while human capital and physical capital contribute 1.8 and 2.2 to the differences in income, respectively

However, neoclassical model treats technological progress as an exogenous factor (that is determined outside the model as a “manna from heaven”) It provides

no insights for policy implications and leaves the driver of long run growth with unexplained manners To overcome limitations of neoclassical growth models, latter economists have developed new growth theory that can explain sources of technology progress as well as the reasons why it affect to economic growth

2.2.2 Endogenous Growth Theory

Endogenous growth theory or new growth theory is an effort of economists (Arrow(1962), Romer (1986), Lucas (1988), Rebelo (1991), and Grossman and Helpman(1991) to explain sources of technology progress Instead of leaving technology progress as unexplained factor, they attempt to explore the channels which technology progress is affected In endogenous growth theory, technology progress occurs through innovation, investment in research and development, etc

2.2.2.1 Learning by doing model of Arrow

Learning by doing concept mentions about ability of workers to improve their productivity through repetition task and practice In learning by doing model, both comparative advantage and growth are involved to trade Trade may change the structure of specialization of a country, and impacts of trade rely on the level of

learning externalities For intra-national spillovers, countries specialize in producing goods with higher potential for learning will grow faster

Learning by doing is first introduced by Arrow (1962) Arrow considers technology progress as a part of economic activities He states that although new knowledge can be produced through repetition task, but it is decreasing In order to continue the learning by doing process, we need to stimulate this process by adding new flow of capital Hence, new investment is considered as a source of leaning by doing We will examine this model comprehensively in the production function take the form of Cobb – Douglas:



 

L BK

where Y is the output of economy, L is the labor force, K is the capital which

is included both physical and human capital B is the level of knowledge increase from learning by doing process

As we mention above, the level of knowledge depend on new flow of capital Hence, we can write the level of knowledge as a function of capital

exogenous factor as in neoclassical theory, and human knowledge is introduced as another form of TFP

2.2.2.2 Research and development model

Paul Romer is the pioneer in the introduction of R&D model As pointed out

by Romer (1990), knowledge or ideas have characteristics of public good, which are non-rival and non-excludable Nevertheless, the use of some specific knowledge can be excluded by legal protection Firms which want to maximize the profits usually engage in doing research, and through patent law they can protect their inventions for a certain period The existence of monopoly profits provides incentive for firms to invest in research and development activities However, it is seem imperfectly to exclude other firms from using the knowledge When the protection time of patent is over, others firms that operate in the same industry can

copy or imitate these knowledge Hence, it is undoubtedly that research activity from a firm also creates positive spillover effect for other firms

We begin with the production function take the form of Cobb-Douglas (see Jones (1995))



) (AL Y K

where Y denotes for output, K is capital, A is stock of knowledge or technology, which is already exist in the economy We can simply understand A as accumulation of all of knowledge that already been created by researchers in the past

This model consists of two sector That is, R&D sector and good sector The particular role of good sector is to produce output for the economy, and R&D sector

is to create new technology or knowledge Labor in the economy is employed to produce goods (LY) or to research and create new ideas (LA) Labor is the economy (L) equal: L = LY + LA Model for R&D sector is:

A

L

A  (2.7) where A denotes for new knowledge or new technology that have just invented,  is the rate at which new knowledge is created This equation shows that new technology (or total factor productivity growth) will increase with the proportion of labor in research activities In addition, we know that the rate which researchers create new knowledge is a function of existence knowledge in the economy and number of scientists (LA) We have:

of knowledge is reduce due to duplication same ideas

Replace 2.7 into 2.8, we arrive at the production function for new knowledge:





It is clear from equation 2.9 that new technology (or knowledge ) depends on amount of scientists and accumulation of knowledge

This equation implies that countries which have bigger stock of knowledge would experience faster total factor productivity growth Second, countries invest more in R&D also has higher TFP growth

Discussion about theories that relates to TFP level and TFPG is very useful

It helps us a clearer understanding about TFP as well as to find the answer for the question “What is the theoretical determinants of TFP?”

2.3 Determinants of TFP level and TFPG

In the neoclassical growth model developed by Solow (1956) and Swan (1956), the fundamental sources of economic growth are capital accumulation as well as technological progress Saving plays an important role in capital accumulation To acquire technology progress, it is essential to have new technology, therefore, changes in technology primarily have a strong impact on TFP

However, the later theories explain the term “TFP” as a measurement of production efficiency Having consider this definition, if any factor affects on input and output relationship, it would have impact on TFP

There are many factors influencing on both TFP level and TFP growth, which is described in endogenous growth theory In addition, it states that four sources of TFP growth, including “economies of scale, resource allocation efficiency, technology progress and human capital” are considered fundamental sources (Huong 2001, p.15) From four fundamental sources, they help economists find out many more factors which affect on TFP growth and TFP level through 4 these important channels such as FDI inflow, export, investment in human capital, research and development, health, infrastructure, institution, technology transfer, etc Some of these factors will be mentioned next

a) Foreign Direct Investment (FDI)

The question whether FDI benefits to productivity of recipient countries is still a controversial question between scholars Many economists believe that FDI is good for productivity growth through technology transfer and technology diffusion channels Motivated by positive expectation of FDI, many developing countries have numerous policies to attract flows of foreign direct investment, and they consider FDI as an important external financing source to boost up economic growth of their country On the other hand, other scholars believe that FDI has no particular impact on productivity growth What is the argument for and against the impacts of FDI on productivity growth? Does FDI really benefit for productivity of recipient countries? In this part, we try to seek the theoretical answers for the these question

When invest and operate in a foreign country, multinational corporations often face with many disadvantages and uncertainties Potential drawback that multinational corporations face is that they lack of understanding about foreign markets and they don’t know much about laws as well as regulations of foreign countries To overcome these disadvantages, multinational corporations must possess world advanced technology, and then it will transfer this technology to its foreign affiliates in order to stay competition Hence, FDI is widely seen as the vehicle for technology transfer, it helps to bring advanced technology into the recipient economy Another point worth noting, FDI also creates positive spillover effects such as knowledge spillovers to domestic firms Domestic firms can learn expertise management knowledge, marketing techniques and advanced production methods from nearby foreign firms through “learning by seeing” process In addition, employees who work in multinational corporations also receives benefits through strictly job training Afterward, they may transfer these advanced knowledge to local firms when they change career or they use these advanced knowledge to set up their own business For example, in an interesting research of Javorcik and Spatareanu (2005) indicates that 25 percent of middle managers in

Czech and 15% percent of managers in Lavia acknowledge that they have study skills and expertise management practices from multinational corporations

Further more, increasing competition between multinational corporations and local firms also makes local firms more productive by employing their resource in a more efficient way to produce better goods and services

Even though there are lots of theories to support the positive effects of FDI

on productivity, results from empirical studies for FDI-productivity nexus are still ambiguous At micro-level study, Aitken and Harrison (1999) find a negative impact of FDI on total factor productivity growth among Venezuelan firms They explain for this negative relationship is due to “competition effect” That is, multinational companies have technological advantages in producing goods and services, so they attract customer’s demand from local firms As a result, local firms have to reduce its production and shift its average cost curve up In addition, Haddad and Harrison (1993) reach a similar conclusion for firms in Morocco

Aitken et al (1996) also conclude that there is existence the negative impact of FDI

on productivity when they conduct a research on Mexico and Venezuela firms At macro level study, Borensztein et al (1998) report a negative relationship between FDI and economic growth of 69 developing countries They point out that recipient countries can only benefit from FDI inflows if recipient countries have sufficient level of human capital Similarly, Nelson and Phelps (1966), and Benhabib and Spiegel (1994) interpret that if stock of human capital is weak recipient countries,

“absorptive capacity” or ability to learn from foreign firms will be limited FDI inflows then may have negative impact on productivity

b) Export

In literature, economists mention about two-way linkage of trade and productivity The first linkage lays stress on important role of export on productivity growth The second refers to reverse linkage from productivity growth

to export However, the pioneers of export-led development theory often emphasize the indispensable role of exports in enhancing productivity and efficiency (Haddad,

De Melo and Horton (1996), and Baldwin (2003)) Export is an important tool to achieve knowledge about production methods through learning by doing and learning by exporting In the model of learning by doing, Arrow (1962, p 155) defines that “Learning is the product of experience Learning can only take place through the attempt to solve a problem and therefore only takes place during activity” He (Arrow, 1962) also states countries specialize in producing and exporting goods with higher potential for learning will grow faster Learning by exporting is a concept to describe the mechanism of enhancement productivity of exporting firms when firms participate in export markets and exploit production knowledge of trading partners That is, in order to export to the foreign markets, firms have to know about foreign customers and their demands Furthermore, when enter into international markets, local exporting firms have to obey rules about quality of product as well as delivery conditions To satisfy these strictly demands, firms usually receive helps from foreign purchasers Foreign purchasing partners will teach firms how to manage production process more efficiently, how to control product quality as well as training workers As pointed out by Grossman and Helpman (1991, p 166) “When local goods are exported, the foreign purchasing agents may suggest ways to improve the manufacturing process” To stay competition in international markets, exporting firms also have to learn advanced technology and apply these technology into production process to produce quality products to meet requirements of foreign customers Therefore, exporting firms could gain benefits from learning by doing and learning by exporting process Further more, by expanding their products to the foreign markets, firms have better chance to obtain greater economies of scale Economies of scale are improved when the production costs are reduced by a bigger in sale volume, and again economies of scale enhance productivity growth Needless to say, there is a variety of convincing empirical researches for why firms operate in exporting sectors are much more productive than non-exporting sectors (see Bernard and Jensen (1999)) Last but not least, thanks to export activities, exporting countries could gain foreign exchange

which is extreme shortage at many developing countries, and then they could import high-tech products and modern machinery This is also one important sources of productivity

Perhaps equally important, there is also another argument to argue the reverse linkage from productivity growth to export Accordingly, exporting firms face with great difficulties and uncertainties when they enter into international markets In this regard, exporters first face with many kinds of trading cost such as cost in searching for potential market, cost in setting up distribution networks and transportation costs, etc They also face with uncertainty about trading in foreign markets They have less knowledge about foreign regulations and they don’t know much about demand of foreign customers than their foreign competitors Because of huge trading cost and uncertainty, only productive firms can bear these cost and engage in export markets Accordingly, Greenaway and Kneller (2007, p.135) assert that “It has become something of a stylised fact that ex-ante productivity determines the choice of whether or not to export In other words, firms have to become more productive before they export and causality runs from productivity to exports Causality in the opposite direction is less clear.” In addition, there is another ideas that support for the reverse linkage from productivity growth to export Clearly, productivity growth of a country will be reflected in its price and its product’s quality That is, if a country is more productive, it will produce the product with lower price and higher quality than other countries Productivity of a country will increase its competitiveness, and again this country could export more product than other countries

If we don’t take into account the reverse causality of export and productivity growth in our model, the results will deliver biased coefficient Hence, we apply instrumental variable techniques to solve endogeneity problems

c) Human Capital

Human capital is the combination of skill, health, experience and knowledge about the production of labor Human capital is considered as a factor which determines technology progress and technology efficiency In the research and development models, Aghion and Howitt (1998) and Romer (1990) find that number of researchers (or human capital) helps to accelerate TFPG through innovative new technology In addition, Nelson and Phelps (1996) provide strong evidences on the important of human capital on TFPG, these evidences reflect the fact that countries with higher level of human capital can easily adopt and implement advanced technology from the technological leader countries

Further more, when evaluating the benefits of FDI inflows in recipient countries, economists often use the term “absorptive capacity” That is, the capacity

of recipient countries in using capital inflows without making a reduction in rate of return of these capital And productivity of the capital inflows will decline, if amount of foreign investment capital employed in production grows faster than skill

or knowledge of workers about the production method (human capital) Borensztein

et al (1998) provides a strong empirical evidence that FDI inflows only benefits for economic growth (or productivity) of recipient countries when recipient countries have enough level of human capital Since in developing countries usually have low level of human capital, thus these countries can not fully exploit the benefits of FDI Hence, FDI may have negative impacts on economic growth (or productivity) in these countries

d) Research and development (R&D)

Economists (Grilliches and Mairesse (1991), Hall and Mairesse (1995)) have widely accepted the positive link between research and development and productivity growth The idea simply is that investment in R&D stimulates innovation Innovation offer great opportunity for innovating firms to reduce production cost as well as enable firms produce new products and services with better quality from existing resources R&D not only provides productivity and

profits for the firms that conduct R&D activity, it also bring benefits for other firms that operate in the same industry through spillover effect Further more, R&D enable domestic countries to develop its absorptive capacity and adapt advanced technology into production in a faster way

e) Health

The connection of health and TFP growth seems to be closely associated Obviously, a healthy workforce will be more productive, and good health will help workers improve their ability to adopt new technology Poor health not only affects

to wealth and income of individuals but also to productivity of the economy Taking malaria as an example, the disease which indicates severe impact of poor health on productivity A person who suffer from malaria usually sick from 12-15 days and he still feel headaches and fatigue after recovery To some extent, malaria is not a fatal diseases, but it lost working time of employee as well as their productivity As a result, labor supply for economy is reduced by poor health Further more, in the viewpoint of foreign investors, they less likely to invest in countries or areas where there is high mortality rates and high disease burden The rationale for their choice

is that workers suffer from illness will have lower working capacity and productivity, and enterprises also bear higher production costs due to increase in hiring and training cost for newly workers in replacing for illness workers who absent from work Hence, some developing countries located in tropical region where suffers from infectious diseases is less likely to receive FDI inflows, and then the potential for improving technology is limited Poor health and diseases also affect to human capital accumulation process through lower school attendance rates

To this end, poor health and diseases may hinder productivity growth by making resources are not distributed in efficient way For instance, in some developing countries, a large government expenditure have to spent on health care system such

as anti-malaria protection as well as fighting against undernourishment Hence, government budget to encourage R&D in private sectors is neglected

f) Institution

Institution has a important role in economic growth and productivity growth According to Acemoglu and Robinson (2010), institution is the main factor to explain for the differences of wealth across countries in the world A good institution can stimulate saving and investment as well as ensure an efficient allocation of resource, and then result in higher TFP growth For instance, Easterly and Levine (2002, cited in Isaksson 2007, p.42) report a situation where political institutions ensure land rights for landholders No doubt, these farmers have incentive to invest in large scale production and enjoy benefit of economics of scale On the other hand, inappropriate institution and policies would have severe impacts on productivity and economic growth Acemoglu, Johnson and Robinson (2001), for example, points out that institutions in colony countries did not secure for private property rights or protect their citizens from risks of expropriation As a result, investment incentives are discouraged Opportunities for investment, innovation and acquiring foreign technology through FDI are also reduced Further more, North (1981), Mokyr (2002), Hall and Jones (1999) , and Ashraf and Galor (2007) have presented evidence that good institutions will facilitate for advanced technological research as well as knowledge diffusion

g) Infrastructure

A good infrastructure system (road, electricity and water supply system) is always needed to foster productivity and efficiency of economy A good infrastructure system helps to encourage investment, capital accumulation as well as technology transfer In their paper, Hall and Jones (1996) state that infrastructure is the key to determine “why some countries have higher levels of productivity, physical capital and human capital than others” In addition, Aschauer (1989, cited

in Isaksson 2007, p 29) finds out that the investment in infrastructure in the US leads to the larger economic return to the society He also states that the downturn

in productivity in the US in 1970s is largely due to a decreasing in public infrastructure investment

2.4 Empirical Studies

Theoretical insights for the relationship between TFP and variables are very necessary and interesting However, empirical evidences that based on theories are more stable and convincing Until now, studies of determinants of productivity and efficiency at country level are still very limited Some research of the others authors are listed below

Table 2.1: Summary of empirical studies relating to the determinants of TFPG

Research Method

Key Findings

TFPG

Export (export to GDP ratio)

Human capital

 Inflation

Term of trade

Outward orientation

Panel data

Use regression analysis with fixed effect to estimate TFP first

After that, find out determinan

ts of TFP

Export shows a positive and statistically significant

at 1% level The contribution

of human capital generally has positive impact on TFP

6 time blocks from 1965-69, 1970-74, 1975-79, 1980-84, 1985-89, and 1990-95

TFPG

Trade openness (ratio of export plus import to GDP)

 Inflation

Water borne diseases

Malnutrition

Malaria

Panel data

Use regression analysis to estimate TFP first

Second, to find

impacts of health variables

on TFPG

Poor health has negative and significant effect on TFP

Research Method

Key Findings

of Pakistan economy

Budget deficit

Government consumption

Population

Education expenditure

Private credit

Employment

Time series data

Use growth accounting method to calculate TFP After that, find out determinants

of TFP

FDI have a positive impact

on TFP growth

of Malaysia economy

TFPG

Investment rate

Trade (ratio

of export plus import to GDP)

Proportion of foreign

enterprise

Education level

Data Envelopment Analysis (DEA) to estimate TFP and finds out possible factors that determine TFP

Open to trade has positive impact

on TFP

Research Method

Key Findings

2000

TFPG

FDI

Human capital

Government share

Initial TFP

of countries compared with

The contribution

of FDI is positive to TFP growth Author doesn’t find evident support the absorptive capability hypothesis

Source: Author’s summarization

In a study of the impacts of openness, trade orientation as well as human capital on TFP, Miller and Upadhyay (2000) use panel data of 83 countries in the

world from 1960 to 1989 By using econometric method (or regression analysis),

the authors first estimate 2 TFP measures from the Cobb-Douglas production

function That is, the authors try to include and exclude human capital into the

aggregate production function as a factor of input Because authors think that it can

create misleading results for OLS estimations without taken into account time

specific or country specific effects, they include 6 dummy variables for 6 time

periods and adjust their data as deviations from the means of specific country over

time Nevertheless, it is worthwhile to note that the calculation of TFP basing on

regression analysis has many problems, and one of the main problems is the

possibility of endogeneity between output and capital as well as output and human

capital As the authors admit that “the reader needs to keep these potential biases in

mind when interpreting our findings” (Miller and Upadhyay (2000), p.8) Second, based on the result of 2 TFP estimations (with and without human capital in production function), they search for main determinants of TFP Of course, they particularly interested in openness, trade orientation, and human capital variables They argue that the more a country trade with the world, the more openness of an economy is, and the greater chance this country can adopt advanced technique of production as well as import key inputs for production They find that the result of a country’s openness has positive and statistically significant on TFP at 1% level Interestingly, trade orientation has robust and negative statistically significant on TFP at 5% level What does it mean when trade orientation has negative sign? Remember that trade orientation is measured by the deviation of domestic price from purchasing power parity When the deviations between the domestic currency and purchasing power parity increase, it means that home currency becomes less undervalued Countries follow policies that lower its real exchange rate below purchasing power parity will have higher TFP Authors also examine the role of human capital to TFP by dividing the data into 22 low income countries, 38 middle, and 23 income countries They find that human capital associate with negative impact on TFP for high income countries, and human capital have positive impact

on TFP for middle income countries For low-income countries, the impact of human capital on TFP will change from negative to positive when these countries have higher degree of openness

In the awareness of the empirical contribution of health on TFP, Cole and Neumayer (2006) follow Miller and Upadhyay (2000) by using panel data of 52 countries from 1965 to 1995 Their paper is unique in discussing the impact of poor health to TFP They argue that poor health affect economic growth of a country through the productivity of labor inputs The researchers suggest three indicators as

a measure for poor health, which is malnutrition (measured by proportion of undernutrion population), malaria ( the proportion of country’s area which is affected by malaria), and water borne diseases (the fraction of population that can

not access to clean water) Authors also address the possibility of endogeneity of malnutrition, malaria, and water borne diseases That is, for example, when level of TFP of a country increase, this country will achieve a higher income Therefore, the proportion of malnutrition will definitely reduce To solve the concern of endogeneity, authors first employ lagged variable of malnutrition, malaria, and water borne diseases in three separate models In order to ensure problem of endogeneity can be ruled out comprehensively, they include three instrumental variables “percentage of population and area of a country in Koppen-Geiger Climate Zone B”, “density of population live in rural and urban”, and “country’s ecology malaria” for variables malnutrition, water borne diseases, and malaria, respectively The paper presents evidents that malnutrition, malaria, and water borne diseases have robust and negative stastiscally significant impacts on TFP Though their empirical work also show evidents about the important of trade to TFP growth

Khan (2006) conducts a study about the determinants of TFP in Pakistan economy By employing time series data from 1960 to 2003, author first utilize the growth accounting method to calculate TFP To get some sense of the factors which affect TFP, he includes variables such as FDI, openness to trade (which is measured

by ratio of import plus export to total GDP), population, etc Note that he classified these above variables into two groups He runs regression with the first group with variables (inflation, budget deficit, education expenditure, openness to trade, financial development, and population) He finds an unexpected result that the relationship between trade and TFP is negative and stastiscally significant With the second group, he adds variables (private credit, domestic investment, employment, consumption of government and FDI inflows)

He finds that FDI have positive and stastiscally significant effect on TFP growth Finally, he takes into account all variables and runs the third regression model The results indicate both trade and FDI are stastiscally insignificant This paper have some problems that need to be taken into account Firstly, the author

employs time series data which have it own problems such as serial correlation and causal relationship between FDI and TFP These problems need to pay close attention, but the authors didn’t Secondly, there are only 43 observations in this study Hence, when the authors added 11 dependent variables, it will loose degrees

of freedom to produce reliable results

In his research, Jajri (2007) examines the contribution of trade, education level (measured by porportion of employees with tertiary education), and foreign ownership on TFP growth on Malaysia The analysis in his paper is carried out in two steps First, he uses Data Envelopment Analysis (DEA) and the Malmquist productivity index to estimate TFPG Second, he figure out possible factors that determine TFPG The author finds that human capital, export have positive effects

on TFP growth

In order to find answer for the question whether FDI has positive effect on TFP growth Woo (2009) uses both cross section and panel data for 92 countries from 1970 to 2000 This paper yield a result that FDI has positive and statiscally significant impact on TFP However, their empirical work in a contradictory way to the absorptive capability hypothesis which state that FDI can foster economic growth of a country when this country has reached a certain level of human capital They don’t find evident that support this hypothesis One problem with this paper

is that there is possibility of endogeneity between FDI and TFPG in cross section model, which requires special attention when we conduct another research

All of the mentioned researches have provided good background for us to conduct our study However, some research has some limitations and they fails to provide an accuracy enough to measure the impact of FDI, export on TFP growth Although some study attempted to estimate the relationship between FDI and technology growth, the relationship between FDI and technology growth is ambiguous FDI inflows can help increase productivity as well as TFP growth through an increase in quality of human capital or organizational know-how, but

countries to be know highly productive is more likely to attract FDI inflows There

is the possibility of endogeneity between FDI and TFP growth

To sum up, both endogenous growth theory and empirical works strongly support the important role of FDI and export on TFP and efficiency According to these literature, FDI is considered as a vehicle to bring advanced technology, tranfer knowledge, boost learning by doing process of recipient country Equally, the contribution of export to TFP should not be neglected Exporting enhance productivity throughout economies of scale, and export help increase efficiency through learning by doing, learning by exporting process

2.5 Conceptual framework

In this section, I will specify the framework as well as variables that relate to

my model For more details of variables in my model, I will present in model specification section

Figure 2.1: Conceptual framework of determinants of TFPG

Specialization

FDI Health

Human Capital

Education

Health

R&D

Technology Progress Innovation

Technology Transfer FDI

Export

Instrumental Variable

Total Factor Productivity

Chapter III RESEARCH METHODOLOGY

Based on theoretical foundation and previous studies about determinants of TFPG, it is the right time to turn our attention to the research methodology which is the central of our empirical analysis This chapter is divided into 3 parts Part 1 is about methods in productivity measurement Next, the sampling method and combination of data will be presented Finally, I will specify models and explain for the choice of variables

3.1 Methods in productivity measurement

Generally, measuring TFPG is currently used in two following methods The first method is the growth accounting exercise which was initiated by Robert Solow (1957) The second method is to use the econometric method or regression analysis

3.1.1 Regression analysis method

In this method, TFP growth is estimated by regressing growth rate of output

on growth rate of input, and TFP growth is interpreted as a residual from aggregate production function However, the estimation of TFP growth by using regression method will introduce various problem One serious problem that has been noted in the theory is the endogeneity of capital and labor As pointed out by The World Bank (2000), “The endogeneity of factor inputs should be considered when assessing the important of TFP growth” This is because in the regression method, there is a correlation of the error term with the inputs in a causal relationship For instance, log of TFP equals to the log output minus the log inputs, but the log of output is clearly correlated with the level of inputs It is explained that countries being more productive will attract both physical capital and labor (human capital), and countries having higher technology will have higher labor (human capital) as well as physical capital investment In addition, authors who use econometric method to estimate TFP growth have to admit “the reader needs to keep these potential biases in mind when interpreting our findings” (Miller and Upadhyay

(2000), p.8) Hence, it is no doubt, Baier, Dwyer and Tamura (2006) has emphasized that employing growth accounting exercise to calculate TFP growth is a more suitable method in comparison with regression analysis In this research, we follow the wisdom advice of Baier, Dwyer and Tamura (2006) by using growth accounting exercise to estimate TFP growth

3.1.2 Growth accounting framework for calculating TFPG

In order to calculate factors like capital, labor, and technological progress contributed to economic growth, I use the Solow growth model - the basic methodology for growth accounting exercise Given the notion that it is impossible

to measure technological progress directly, the growth rate of technology is measured indirectly by calculating the different between the actual growth rate of output (GDP) and the part of growth rate of capital as well as growth rate of labor

Starting from a production function takes the form of Cobb-Douglas:



Yit  AitKitLit (3.1) where Yit stands for total output (GDP) of country i (i=1, 2, 3…103) in year t, Kit is the capital stock in country i in year t, and Lit is the quantity of labor in country i in year t Ai is the level of TFP in year t for country i  is capital share of output and

β labour share of output

Taking logs and differentiating both sides with respect to time, we arrive at the following formula for TFP growth over two years:

i i i



1 Following the tradition in the neoclassical literature going

back to Solow (1956),  and β can also be measured as the income shares of capital and labor

The income share of labor (β) can be calculated as the ratio between the wage of employees and the Gross Domestic Product (GDP) for each country in the 1996-2009 period Thus, the capital share in total output () is calculated by  = 1-

β

However, data for income share of labour as well as income share of capital

is overwhelming and hard to collect for individual countries, almost of researches (Baier, Dwyer and Tamura (2006), Woo (2009)) assume a constant income share of labor for all countries and years In addition, Gollin (2002) finds strong evidence to support the hypothesis of common income share of labor for all countries and years, and he finds that the income share of labour for cross country usually ranges from 0.6 to 0.8 The average number that most of researchers usually choose equal 0.65 Further more, Woo (2009) uses both constant income share of labor and actual income share of labor data from Bernanke and Gurkaynak (2001, cited in Woo

2009, p.229) to estimate TFP growth Again, he finds very similar results of TFP growth from constant income share and actual income share Hence, the use of constant income share of labor is not a big problem in our paper We follow Woo (2009) by using constant income share of labor equal 0.65 for all countries and years Thus, income share of labor equal 0.35 To some extent, we strongly believe that the use of constant income share of labor equal 0.65 is also suitable with our data set since Woo (2009)’s data set include both developed and developing countries from 1970 to 2000 And there is also empirical evidence to support the hypothesis of constant income share of labor for all years

3.2 Data source

The secondary data is mainly used in this study The data is the aggregate data of 103 countries in the world (including Vietnam) from 1996 to 2009 The reason for choosing 103 countries from 1996 to 2009 as our sample reflects the availability of data that we can collected 103 countries is the biggest number of countries that we have enough data to estimate TFPG continuously from 1996 to

2009 For more information about countries in our sample, please refer to Table A1

on Appendix 1 Beside, I use data from 1996 to 2009 to study since this is the most recent data which would have higher quality than previous period, and it will give the most up-to-date picture about the effects of FDI and export on TFPG It is important to notice that our sample also captures the impact of two serious financial crisis in this period, that is, Asian financial crisis (1997) and global financial crisis (2008) We argue that if the results is significant in this period, then it should be significant even stronger for others period

3.2.1 Data for dependent variable TFPG

Clearly, there is no data for TFPG for 103 countries in the world between

1996 and 2009 period In order to study impacts of FDI and export on TFPG, we have to estimate TFPG by employing growth accounting exercise as mentioned above To calculate TFPG, we need data on output (GDP), capital stock input and labor input

• In my paper, I use real GDP as an indicator for output growth Real GDP data of each country can be collected from World Development Indicators of World

Bank GDP data is deflated and measured in 2000 constant USD

http://databank.worldbank.org

• To measure the labor input, I collect data on total employment for each

country Data on total employment of each country can be collected from World

Development Indicators of World Bank http://databank.worldbank.org

• Data on capital stock input is not available for most of the countries in the sample Again, I have to calculate capital stock for 103 countries I apply “The perpetual inventory method” The perpetual inventory method has the following formula:

Kt = (1 – δ)Kt-1 + It (3.3) where Kt is the capital stock in year t, Kt-1 is the capital stock in year t-1

To compute the capital stock, we need data on depreciation rate and initial capital stock level (K0) In our sample, K0 is the year 1996 I follow the method of

Hall and Jones (1999) in the construction of initial capital stock by using the following formula:

I

g

I K

Furthermore, with the assumption that capital stocks are homogeneous, it means that all capital stock have the same depreciation rate among group of countries over time For the value of depreciation rate δ, we use results from Bu (2004) That is, depreciation rate equal 5 % for developed countries and equal 7 % for developing countries Bu (2004) points out that developing countries always experience higher depreciation rate than developed countries This is because developed countries possess a better maintenance system than developing countries

It is the value of total investment Data of It is available at World Development Indicators of World Bank http://databank.worldbank.org Data of total investment It is deflated in 2000 constant USD

3.2.2 Data for independent variables

In this section, I only mention about the source of data which I collected from For more information about the usage of data, I will present in model specification section

• Export is measured by ratio of export to GDP (%) Data of export for 103 countries is available at http://databank.worldbank.org

• FDI is measured by ratio of net FDI inflows to GDP (%) Data of FDI inflows are colleted from the UNCTAD database It is available at http://www.unctad.org/Templates

• The main source of inflation data is obtained from World Development Indicators

of World Bank http://databank.worldbank.org Inflation is proxied by annual percentage of consumer price index (%)

• Human capital (HC) data is obtained from World Development Indicators of World Bank http://databank.worldbank.org and Barro and Lee Education dataset http://www.barrolee.com/data/dataexp.htm Human capital data from both of the source is proxied by average secondary schooling years of population over 15 years old (%) In cross section study, we employ human capital from World Development Indicators of World Bank In panel data exercise, we use the data from Barro and Lee education dataset The rationale for employing the data from Barro and Lee education dataset in panel data is that the data set has data on a five year period base for each country It is suitable with my panel data because we also use data of five year average for each country, and we also want to check the robustness of FDI and export when we employ different data source for human capital

• Data of population is collected at Penn World Table 7.0 It is available at http://pwt.econ.upenn.edu/

• We also employ government expenditure in our model Government expenditure

is measured as ratio of government consumption to GDP (%) Data of Government expenditure is collected at Penn World Table 7.0 It is available at http://pwt.econ.upenn.edu/

• Data of land area (km2) and data of distance from latitude (km2) that will be used for instrumental variables is collected from Center for International Development It

is available at: http://www.cid.harvard.edu/ciddata/geographydata.htm#general

• Data of landlocked economies come from CEPII center It is available at

http://www.cepii.fr/anglaisgraph/bdd/distances.htm

3.3 Model Specification

I this section, I will present models for both cross section and panel data Firstly, I calculate TFP growth from 1996-2009 for each country using growth accounting method, and it will create a cross section data set of TFP growth along with variables such as FDI, export, human capital, etc from 1996 and in 2009 Secondly, as a more comprehensive method for robustness check, I also report the