The influence of Flexible Manufacturing Technology adoption on ...

Economic Modelling 27 (2010) 395–403

Contents lists available at ScienceDirect

Economic Modelling j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / e c m o d

The influence of Flexible Manufacturing Technology adoption on productivity of Malaysian manufacturing industry D.A.R. Dolage a, Abu Bakar Sade b, Elsadig Musa Ahmed c,⁎ a b c

International Graduate School of Business, University of South Australia, Adelaide, Australia Faculty of Management, Multimedia University, Cyberjaya, Malaysia Faculty of Business and Law, Multimedia University, 75450 Melaka, Malaysia

a r t i c l e

i n f o

Article history: Accepted 9 October 2009 Keywords: Flexible Manufacturing Technology Total factor productivity growth Mass customisation

a b s t r a c t This paper investigates the influence of the adoption of Flexible Manufacturing Technology (FMT) on the Total factor Productivity Growth (TFPG) of Malaysia Manufacturing Industry. The Principal Component Analysis has been adopted to extract the most appropriate underlying dimensions of FMT to use in place of the eight FMT variables owing to the potential multicollinearity. The study has been conducted within FMT intensively adopted 16 three-digit industries that encompass 50 five-digit industries covering the years 2000–2005. The results obtained from the two situations, one, including the industry fixed effects dummy variables and the other without these, are contrasted. It is established that the model that included the industry fixed effect dummy variables has a greater explanatory power. The two principal components that account for the greater variation in FMT show positive and moderately significant relationship with TFPG. The study provides sufficient evidence to conclude that FMT has a direct and moderately significant relationship with TFPG. © 2009 Elsevier B.V. All rights reserved.

1. Introduction The average gross domestic product (GDP) growth of Malaysia (5.5%) during 2000–2007 is lower than that (7.0) during 1990–2000. Malaysian Manufacturing sector GDP (13.0%) during 2000–2007 is much lower than the same (4.8%) for the period 1990–2000. These are some of the key indicators to the declining competitiveness of the Malaysian manufacturing industry over the period 2000–2007. It is widely believed that intensive regimes of contemporary manufacturing paradigms such as mass customisation, customerisation and instant customerisation can pave the way for a competitive manufacturing industry. The studies show that mass customisation is the core manufacturing paradigm. The studies also showed that the crucial determinant of the successful implementation of mass customisation is the abundant use of Flexible manufacturing Technology (FMT) (Wind and Rangaswamy, 2001; Da Silveria and Fogliatto, 2005). Moreover, Malaysian Industrial Development Authority (MIDA) (MIDA, 2007) has recognised a number of promoted activities and products (for the development and production) for high technology establishments which makes them entitled to pioneer status or investment tax allowance under the promotion of Investment Act 1986. This includes FMT products such as, Computer process control ⁎ Corresponding author. Tel.: +60 126330517, +60 62523807; fax: + 60 62318869. E-mail addresses: [email protected] (D.A.R. Dolage), [email protected], [email protected] (A.B. Sade), [email protected], [email protected] (E.M. Ahmed). 0264-9993/$ – see front matter © 2009 Elsevier B.V. All rights reserved. doi:10.1016/j.econmod.2009.10.005

systems/equipment, Process instrumentation, and Robotic equipment and Computer numerical control machine tools. The Ninth Malaysia Plan which is compiled by the Economic Planning Unit of the Prime Minister's Office presents the first five-year blueprint of the National Mission, outlining the policies and key programmes aimed at fulfilling the Mission's ‘Thrusts’ and objectives for the period 2006–2010. This aims to achieve changes in the structure and improved performance of the economy with every economic sector achieving higher value added and total factor productivity. The ‘Thrust 1’ of the Plan is aimed at making the economy more centred on human capital, particularly with increasing competition from globalisation and progressive market liberalisation. This states that, ‘Application of high technology and production of higher value added products will be given emphasis. Measures will be undertaken to migrate the electrical and electronics (E&E) industry towards hightechnology and higher value added activities’. The empirical studies on FMT are clustered in the following areas; types of flexibility, types of FMT, procedure bias on investment appraisal of FMT, operational problems, market structure and competitiveness. Nonetheless, it is observed that the influence of FMT adoption on the competitiveness of the Malaysian manufacturing industry has not been adequately explored. Studies have revealed that due to the potential operational problems of FMT implementation, potential benefits of FMT might not be derived (Sharma, 2002; Gale et al., 2002; Roller and Tombak, 1993). Moreover, Slagmulder and

396

D.A.R. Dolage et al. / Economic Modelling 27 (2010) 395–403

Bruggeman (1992) and Fine and Freund (1990) showed that due to ‘procedure bias on investment appraisal of FMT’, investments in FMT do not take place smoothly or effectively. Hence, additional studies need to be carried out to measure the extent to which FMT contributes to the productivity in the manufacturing industry. Evidently, only a few studies have examined the impact of specific technologies on the industry level productivity using less aggregated data. Berndt and Morrison (1995) examined the impact of high-tech investments on multifactor productivity (MFP) and three profitability measures. While the study found only limited evidence of a positive relationship between profitability and the share of high-tech capital in the total physical capital stock, it established that they were negatively correlated with MFP. Amato and Amato (2000) have investigated the impact of high-tech investments on MFP and Price Cost margin. This study established that there was a positive impact from high-tech investments regardless of whether or not the specification includes industry effects dummy variables to account for the differences in technological opportunity among industries. According to Henricsson and Ericsson (2005), Wysokinska (2004) and Porter (1990), productivity is the only relevant measure of competitiveness. Zhi et al. (2003) showed that there were three productivity concepts currently adopted to measure the productivity in the manufacturing industry. Chau (1993) and Oulton and O' Mahony (1994) emphasised on the unresolved academic debate over whether MFP and Total Factor Productivity Growth (TFP) are the same. On account of this, the present study adopts a widely recognised productivity measure, growth of TFP (TFPG). A number of studies have been conducted in the manufacturing industry of Malaysia that adopted TFPG to measure productivity at industry level. Menon (1998) studied TFPG of foreign and domestic firms in the Malaysian manufacturing industry. Tham (1997, 1995), Choong and Tham (1995) and Fatimah and Mohd (2004) adopted TFPG to examine the influence of trade policies and industry characteristics on the productivity growth of the Malaysian manufacturing industry. Abdullah and Hussein (1993) adopted TFPG to examine the productivity growth of the Malaysian resource based industries. These studies indicate that TFPG has been used in Malaysia to measure productivity growth in the manufacturing industry. Moreover, Elsadig (Elsadig, 2006a, 2006b, 2006c, 2007, 2008a, 2008b), estimate TFPG contribution to Malaysia's manufacturing in relation to input driven, positive and negative externalities, such as the impact of information and communications technology, human capital, foreign direct investment, carbon dioxide emissions and Biochemical Oxygen demand emissions. The purpose of this paper is to examine the influence of FMT adoption on TFPG in selected manufacturing industries of Malaysia. This adds to the previous literature by focusing more narrowly on the influence of adoption of FMT on productivity. This study developed inclusion criteria and selected FMT intensively adopted 16 MSIC three-digit industries and 50 MSIC five-digit industries included within them. All secondary data required for the study came from the Annual Surveys of Manufacturing Industries (ASMI) during 2000– 2005 and Economic census data maintained by the Department of Statistics Malaysia (DOS). Another novelty in this study is that prior similar studies have been carried out at the four-digit level whereas the present study is carried out at five-digit level. The present study contributes to the previous studies by considering less aggregated data and also by considering TFPG in place of MFP. This study also considers a higher number of specific FMT variables such as, Computer Numerical Control machine tools, Numerical Controlled Machine Tools (NC), Robotics (ROB), Programmable Logic Controllers (PLC), Automated Inspections (INS), Automated Storage and Retrieval Systems (ASR), Computer Aided Design (CAD) and Local Area Networks (LAN). In order to overcome multicollinearity among FMT variables, the study extracts three underlying dimensions of FMT by adopting Principal Component Analysis. They are namely; ‘process control’ technologies, ‘production and quality control’ technologies

and the ‘general control’ technology. The study adopts a questionnaire survey to compute the degree of adoption of FMT among the selected 50 five-digit industries. The present study considers eight types of FMT instead of five specific technologies, evidently the maximum number considered in a prior study. The study covers only six years from 2000 to 2005 due to the limitation of data availability. The fact is that the DOS follows the MSIC 2000 in classifying industries for the collection and publication of data. The Annual Survey of Manufacturing Industries (ASMI) reports from 2000 onwards have been prepared according to this classification and up to 1999 according to the older version of the MSIC. The older classification system is so different that more than 30% of the MSIC five digit industries (classified according to MSIC 2000) considered in this study is neither listed nor coded or described differently in ASMI reports published up to 1999 which were based on older classification system. Hence, the earliest year that was considered for this study is 2000. Since ASMI 2006 which publishes data for the reference year 2005 was released in early June 2008, the latest year considered is 2005. This has been done by Amato and Amato (2000) too in their study on impact of high-tech investments in profitably and productivity have considered only five years. Since the data for six year have been reviewed the total number of resultant observations (cases) available for this study was 300 (6 × 50 = 300). 2. Methodology The basic research hypothesis of the study is: a high degree of FMT adoption enhances TFPG of the manufacturing industry of Malaysia. TFP is measuring the relationship between output and its total inputs (a weighted sum of all inputs), by this means giving the residual output changes not accounted by total factor input changes. Being a residual, changes in TFP are not influenced by changes in the various factors which affect technological progress such as the quality of factors of production, flexibility of resource use, capacity utilisation, quality of management, economies of scale, and the like (Rao and Preston, 1984). In addition, it has been documented in empirical work on economic growth by Solow (1956, 1957), that after accounting for physical and human capital accumulation, “something else” accounts for the bulk of output growth in most countries. Both physical and human capital accumulations are certainly critical for economic growth. The process becomes more complicated with the role of knowledge in the economic growth process. Knowledge obviously accounts for a part of the growth that is not accounted for by the other factors of production; namely capital and labour. In growth theory, the Solow residual is an unexplained residual of labour and capital and it is attributable to the growth of TFP. The notion of TFP is interpreted as an “index of all those factors other than labour and capital not explicitly accounted for but which contribute to the generation of output.” TFP refers to the additional output generated through enhancements in the efficiency accounted for by such things as advancement in human capital, skills and expertise, acquisition of efficient management techniques and know-how, improvements in an organisation, gains from specialisation, introduction of new technology, innovation or upgrading of present technology and enhancement in Information and Communication Technology (ICT), (Elsadig, 2006a,b,c, 2007, 2008a,b). 2.1. Estimation of TFPG There are two stages in the methodology. The first stage is to estimate TFPG for all the industries considered in the sample. The second is to identify the explanatory variables of TFPG. The TFPG approach to measuring productivity is widely used in the manufacturing industry (Sharma, 2002; Bloch and Tang, 1999; Leung, 1997; Yean, 1997). Jorgenson et al. (1987), based on the pioneering works of


Solow (1957) and Denison (1967) developed an approach to compute TFPG. This measure is based on the Translog (Transcendental Logarithmic) production function which is a generalisation of the Cobb–Douglas production function. According to Zhi et al. (2003), this is based on the premise that for each industry, there exists a transcendental logarithmic (translog) production function giving output as a function of intermediate input, capital input, labour input and time. In this approach, TFPG of each industry is computed as the difference between the rate of growth of output and the average weighted growth of intermediate input, capital input and labour input. The average productivity growth term V′T represents the Translog index of TFP growth i.e. Δln (TFP). This study adopts the same equation adopted by Zhi et al. (2003) to compute V′T which is given below: V’T = ½ln ZðTÞ− ln ZðT−1Þ–V’X ½ln XðTÞ− ln XðT−1Þ

ð1Þ

–V’K ½lnKðTÞ− lnKðT−1Þ–V’L ½lnLðTÞ− lnLðT−1Þ Where VX’ = 1 =2 ½VX ðTÞ + VX ðT−1Þ VK’ = 1 =2 ½VK ðTÞ + VK ðT−1Þ VL’ = 1 =2 ½VL ðTÞ + VL ðT−1Þ VX = pX X = qZ; VK = pK K = qZ; VL = pL L = qZ In the above equations, V′X, V′K and V′L represent the respective shares of inputs of intermediate, capital and labour, averaged over time T and T − 1. While q, pX, pK and pL denote the prices of the output and inputs of intermediate, capital and labour respectively, Z represents the output. Since Δ lnðTFPÞ = V’T TFPðTÞ = V’T ln TFPðT−1Þ TFPðTÞ −1 = ln VT′ TFPðT−1Þ But; TFPG =

TFPðTÞ−TFPðT−1Þ TFPðT−1Þ

Hence; TFPG =

TFPðTÞ −1 −1 = ln VT′ −1 TFPðT−1Þ

2.2. Review of factors affecting TFPG in the manufacturing industry Growth Rate of Output (GRO): it is logical to anticipate that output growth can lead to higher TFPG because it allows ‘economies of scale’ to be exploited. As output grows, capacity utilisation is bound to increase leading to a fall in average cost. A positive link between output growth and TFPG, known as Verdoorn's Law named after P. J. Verdoorn, was postulated during the interwar period (Sharma et al., 2000). Goldar & Kumari (2003), Zhi et al. (2003), Sharma et al. (2000), Amato & Amato (2000) and Leung (1997) considered GRO as an explanatory variable of TFPG. Hence, a positive link is expected between TFPG and GRO, ceteris paribus. Change in Capital Intensity (CCAPIN): it is rational to expect that capital-intensive industries offer more scope for technological progress and learning by doing and thereby influencing the factor productivity. The studies of Goldar & Kumari (2003), Sharma et al. (2000), Leung (1997), Yean (1997) and McGuckin and Streitwieser (1996) revealed that CAPIN was an explanatory variable of TFPG. Amato & Amato (2000) have taken CAPIN as a control variable in their investigation into MFP of high-tech investments. However, the efficiency of capital intensity is more likely to depend on the

397

availability of efficient infrastructure. Sharma et al. (2000) and Leung (1997) in their studies assumed a negative relationship between CAPIN and TFPG. A significant variability could be observed among the researchers in their approach to quantifying CAPIN. Sharma et al. (2000) defined it as Fixed Capital divided by Total Employment. Goldar & Kumari (2003) used the ratio of Investment to Capital. Leung (1997), McGuckin and Streitwieser (1996) and Mahadevan (2002) used the capital labour ratio, measured as the capital per employee. Amato & Amato (2000) defined capital intensity as the value of shipment divided by capital stock. The present study adopts value of assets divided by total number of employees as the capital intensity. The CCAPIN was (measured as the change of capital intensity between two consecutive years) considered in place of its absolute value since TFPG was a variance associated with two consecutive years. A positive link is expected between TFPG and CCAPIN, ceteris paribus. Export Growth (EXPGROW): export growth can lead to higher productivity due to a number of reasons namely; opportunity for greater capacity utilisation particularly in industries in which the minimum efficient size of plant is large relative to the domestic market, greater horizontal specialisation as each firm concentrates on a narrow range of products, and increasing familiarity and absorption of new technologies (Sharma et al., 2000). Moreover, with foreign exchange earned from export growth, firms would have better access to imported inputs and new technology the effects of which can evidently enhance factor productivity. Sharma et al. (2000), Leung (1997) and Yean (1997) considered export intensity or growth as an explanatory variable of TFPG. Since the dependent variable, TFPG is associated with a growth; it is preferable to use EXPGROW as the explanatory variable. This was measured as the increase in exports in two consecutive years divided by the exports of the previous year. A positive link is expected between TFPG and EXPGROW, ceteris paribus. Change in Firm Size (CFSIZE): this variable is a crude proxy for scale of entry barrier. Theoretically, the minimum efficient plant size is a better proxy but could not be included due to the non availability of data. Higher productivity gains can be expected in the presence of oligopolistic competition. Therefore, researchers include average plant size or firm size in TFPG models to take account of such effects: (Chandrasiri, 2003; Amato & Amato, 2000; McGuckin and Streitwieser, 1996). In the present study, FSIZE was measured as the average firm size of the eight largest firms in each industry. A positive link is expected between TFPG and CFSIZE, ceteris paribus. Change in Industry Concentration (CCR4): oligopoly theory explains that the higher the level of concentration, the more likely it is that the dominant firms will be able to collude, tacitly or expressly, to rise prices above the long run average costs (Shiraz, 1973). Therefore, it is reasonable to include this variable in TFPG model as it can affect the productivity in a given industry. Industry concentration is widely expressed in terms of four-firm concentration ratio (CR4) i.e. sales of the four largest firms divided by the total sales in an industry. Amato and Amato (2000) have incorporated CR4 in its MFP model. In the present study, CR4 was measured as the percentage of industry sales contributed by the four largest firms in each MSIC five-digit industry for each year. A positive link is expected between TFPG and CCR4, ceteris paribus. The eight types of FMT considered in this study are given below: Computer Numerical Control Machine Tools (CNC): measured as the percentage of firms in each MSIC five-digit industry using microprocessor based numerical control technologies referred to as computer numerical control machine tools. Numerical Controlled Machine Tools (NC): measured as the percentage of firms in each MSIC five-digit industry using numerical controlled machine tools. Robotics (ROB) measured as the percentage of firms in each MSIC five-digit industry using robotics.

398


Programmable Logic Controllers (PLC): measured as the percentage of firms in each MSIC five-digit industry using programmable logic controllers. Automated Inspections (INS): measured as the percentage of firms in each MSIC five-digit industry using automated sensor-based inspection, either during the production process or final product. Automated Storage and Retrieval Systems (ASR): measured as the percentage of firms in each MSIC five-digit industry using automated storage and retrieval systems. Computer Aided Design (CAD): measured as the percentage of firms in each MSIC five-digit industry using computer aided design to control manufacturing machinery. Local Area Networks (LAN): measured as the percentage of firms in each MSIC five-digit industry using local area networks. A positive link is expected between TFPG and any type of FMT variable, ceteris paribus. Industry Fixed Effects Dummy Variables (INDj): The study involved 50 five-digit industries included in 16 three-digit industries. It is logical to assume that industry characteristics among these 16 three-digit industries can be diverse and need to be captured by a variable. Therefore, 16 dummy variables (INDj) were incorporated into the TFPG model to capture the industry fixed effects. However, most contemporary researchers have not considered industry fixed effects in the TFPG model. While Mahadevan (2002) and Amato & Amato (2000) have incorporated industry fixed effect dummy variables, Goldar & Kumari (2003), Tham (1997), Leung (1997) and Sharma et al. (2000) in their similar studies, have not made any reference to industry fixed effects, let alone considering them in their models. A positive link is expected between TFPG and INDj, ceteris paribus. The model representing the relationship among TFPG, explanatory variables and FMT variables can be specified as given below: TFPG = α0 + α1 GRO + α2 CCAPIN + α3 EXPGROW + α4 CFSIZE + α5 CCR4 + α6 CNC + α7 NC + α8 ROB + α9 PLC + α10 INS + α11 ASR + α12 CAD + α13 LAN + Σα13 + j INDj + μ

ð2Þ

The core hypothesis tested in this research study is: a high degree of Flexible Manufacturing Technology adoption enhances Total Factor Productivity Growth of the manufacturing industry of Malaysia. 3. Data and estimation and inclusion criteria According to the Malaysian Standard Industrial Classification 2000 there are 53 three-digit industries. In order to obtain a rational outcome, the study needs to be conducted only within industries in which FMT is intensively adopted. On account of this, inclusion criteria were formulated in an effort to select FMT intensively adopted MSIC three-digit industries for the sample, which is shown below: Industries with high ‘capital/labour’ ratio Industries in which product variation is a marketing strategy Industries in which products are susceptible to demand fluctuation Using the above criteria a sample of 16 MSIC three-digit industries which together comprise 50 five-digit industries was selected. 3.1. Primary data The data that indicate the degree of adoption of FMT is not published by any organisation in Malaysia. Hence, a questionnaire survey was conducted to gather information necessary to compute

the percentage of establishments adopting each specific type of FMT in a given year, within a given MSIC five-digit industry. The questionnaires were sent to all the establishments, listed under the 50 MSIC five-digit industries that appeared in the directory of Federation of Malaysian Manufacturers. 3.2. Secondary data In order to compute TFPG, industry-wise data is required for output, intermediate input, capital input and labour input. The closest indicators for these values were obtained from Table 3 of the ASMI published for the years 2000 through 2005 by the DOS of Malaysia. The variables GRO and CCAPIN were computed using the data obtained from Table 3. EXPGROW, CCR4 and CFSIZE were computed using the data obtained from the Economic Censes conducted by the DOS Malaysia. 4. Empirical results In view of the fact that only FMT intensively used industries were included in the sample, naturally some similarity in the sequence and characteristics of the production processes could be expected even amongst different five-digit industries. Hence, there could be a tendency for a similarity in the technology adopted amongst these industries. Due to the similarities in technologies, a high prevalence of multicollinearity among the eight types of FMT could be anticipated. In this study, bivariate Pearson product-moment correlation analysis has been conducted using SPSS to test for multicollinearity amongst FMT. The output that reveals potential multicollinearity among FMT variables is displayed in Table 1. According to Coakes et al. (2008) and Field (2005), when a considerable number of correlations are exceeding 0.3, the matrix is suitable for Principal Component Analysis (PCA). PCA was performed using SPSS in order to obtain underlying dimensions (Principal Components) of FMT as a remedy for multicollinearity. As per both standard methods of (i.e. screen test and Eigen values greater than one) extracting the optimal number of components, three Principal Components (PCs) were extracted that account for 67% of the variation in the FMT. According to Table 2, the loadings of variables onto the three PCs obtained from both types of rotations (Orthogonal and Oblique) are quite similar. Hence, due to simplicity, PCs obtained from orthogonal rotation was used in the rest of the analysis. Once the most appropriate type of rotation and the resultant PCs were decided, the variables loading onto each of these PCs were examined as the next step. An examination of the component loadings depicted in Table 2 indicates that Local Area Networks, Computer Aided Design, Programmable Logic Controllers and Computer Numerical Control Machine Tools load onto PC1; Automated Storage and Retrieval Systems, Automated Inspections and Robotics load onto PC2 while only Numerical Controlled Machine Tools load onto PC3. Usually, it is difficult to give clear cut themes or names to PCs that only relate to or encompass particular variables that are loading onto it. Hence, only the best possible names have been assigned to the PCs extracted from this analysis. The technologies Local Area Networks, Computer Aided Design, Programmable Logic Controllers and Computer Numerical Control Machine Tools are used in the manufacturing set up as process control technologies. Since these load onto PC1, so can be named as ‘process control’ technologies. The technologies, Automated Storage and Retrieval Systems, Automated Inspections and Robotics load onto PC2, so can be named as ‘production and quality control’ technologies. PC3 has only one variable i.e. Numerical Controlled Machine Tools, loading onto it so can be called the ‘general control’ technology. As the next step, the eight FMT variables were substituted with the three PCs namely, PC1, PC2 and PC3. Therefore, the TFPG model was


399

Table 1 Correlations among FMT.

CNC

NC ROB PLC INS ASR CAD LAN a b

Pearson correlation Sig. (2-tailed) N Pearson correlation Sig. (2-tailed) Pearson correlation Sig. (2-tailed) Pearson correlation Sig. (2-tailed) Pearson correlation Sig. (2-tailed) Pearson correlation Sig. (2-tailed) Pearson correlation Sig. (2-tailed) Pearson correlation Sig. (2-tailed)

CNC

NC

ROB

PLC

INS

ASR

CAD

LAN

1.000

.160a .005 300 1.000

.351a .000 300 .012 .836 1.000

.634a .000 300 .164a .005 .368a .000 1.000

.307a .000 300 .177a .002 .250a .000 .302a .000 1.000

.237a .000 300 .126b .030 .427a .000 .257a .000 .564a .000 1.000

.248a .000 300 .141b .014 .391a .000 .394a .000 .115b .046 .308a .000 1.000

.322a .000 300 .171a .003 .236a .000 .380a .000 .186a .001 .129b .025 .609a .000 1.000

300 .160a .005 .351a .000 .634a .000 .307a .000 .237a .000 .248a .000 .322a .000

.012 .836 .164a .005 .177a .002 .126b .030 .141b .014 .171a .003

.368a .000 .250a .000 .427a .000 .391a .000 .236a .000

.302a .000 .257a .000 .394a .000 .380a .000

.564a .000 .115b .046 .186a .001

.308a .000 .129b .025

.609a .000

Correlation is significant at the 0.01 level (2-tailed). Correlation is significant at the 0.05 level (2-tailed).

reformulated as follows (the changes in PCs are considered here to be consistent with TFPG which too is measured as a change): TFPG = α0 + α1 GRO + α2 CCAPIN + α3 EXPGROW + α4 CFSIZE + α5 CCR4 + α6 CPC1 + α7 CPC2 + α8 CPC3 + Σα8+ j INDj + μ

ð3Þ

4.1. Multiple regression analysis of TFPG As described, the model contains a set of 16 industry fixed dummy variables (INDj) to account for the differences of technological opportunity among industries. Although it is theoretically desirable to include INDj, the consequent impact of adding these 16 extra variables needs to be examined by comparing and contrasting the results obtained without considering the INDj in the model. A separate regression was performed for this scenario and the tables of Model Summary, ANOVA and Coefficients were obtained. In order to facilitate easy comparison of the results, the tables of output obtained from regression analysis for the two situations, one with the INDj included and the other without the INDj have been combined into one. The tables of ANOVA and Coefficients contained in the SPSS output for these situations have been reproduced in Tables 3 and 4 respectively. According to Table 4, Adjusted R square is considerably high (0.518). This indicates that the explanatory variables together explain 51.8% of the variance in TFPG. However, the explanatory power of the model has decreased marginally when the INDj has been excluded; Adjusted R square (0.50) has decreased.

According to ANOVA, the F statistics for both situations of including and excluding INDj in the models are 14.089 and 36.339 respectively. They both are larger than the critical value (1.53) of the F distribution, obtained from the F distribution calculator for α = 0.05 level of significance when degrees of freedom are 23 and 257. One of the assumptions of regression analysis is that the residual terms should be uncorrelated for any two observations. Since the study involved time dependent variables, the lack of autocorrelation has to be tested. This can be tested with Durbin–Watson Test which tests for serial correlations between errors. According to Field (2005), the test statistic can vary between 0 and 4 with a value of 2 meaning that the residuals are uncorrelated. Table 4 depicts Durbin–Watson test statistics of 2.118 and 2.086 respectively when industry fixed effect variable (INDj) is included and excluded. Since these are very close to 2, it can be concluded that the error term does not show any appreciable autocorrelation. The statistical test for the existence of a linear relationship between dependent variable and the independent variables is: H0

α1 = α2 = α3 = …….. αk = 0

H1

Not all the αi (i = 1,2,……24) are zero

As the F statistic is in the rejection region, H0 was rejected and H1 was accepted. Since ‘p < 0.000’, it can be concluded that there is strong evidence of TFPG having a linear regression relationship with any of the explanatory variables in the model with a probability of less than 0.1% of making an error in this conclusion.

Table 2 Comparison of components obtained from two types of rotations. Component One Orthogonal

LAN CAD PLC CNC ASR INS ROB NC

.816 .816 .666 .555

.477

Component Two Oblique

Orthogonal

Pattern

Structure

.861 .841 .640 .517

.811 .801 .722 .621

.412

.542

.845 .816 .526

Component Three Oblique

Orthogonal

Pattern

Structure

.858 .844 .460

.445 .467 .851 .826 .573 .883

Oblique Pattern

Structure

.871

.883

400


Table 3 ANOVA. Sum of squares

Regression Residual Total

df

Mean square

F

Sig.(p-value)

INDj inc

INDj exc

INDj inc

INDj exc

INDj inc

INDj exc

INDj inc

INDj exc

INDj inc

INDj exc

45.333 36.127 81.68

42.198 39.482 81.680

23 257 280

8 272 280

1.981 0.141

5.275 0.145

14.089

36.339

0.00

0.00

Dependent variable: TFPG, inc — included, exc — excluded.

4.2. Industry fixed effects dummy variables (INDj) included According to Table 4, the variables namely, GRO (0.000) and EXPGROW (0.000) are very highly significant at ‘p < 0.001’. This implies that the chances of making an error by assuming that these correlate with TFPG are less than 0.1%. Also both these variables are positively correlated. CCAPIN (0.009) and CCR4 (0.010) are highly significant at ‘0.001 < p < 0.01’. CFSIZE (0.10) is moderately significant at ‘0.05 < p < 0.1’. While CCR4 shows a positive relationship, CCAPIN and CFSIZE show a negative relationship with TFPG. Since the main focus of this model is to test the significance of the correlation of FMT with TFPG, an examination of the correlation of the three PCs with TFPG becomes necessary. CPC1 (0.034) at ‘0.01 < p < 0.05’ shows a significant and positive correlation with TFPG whereas CPC2 (0.114) at ‘0.10 < p < 0.15’ shows a marginally significant and positive relationship with TFPG. However, CPC3 (0.473) shows highly insignificant and negative relationship with TFPG. Table 3 shows the results of the estimating equation obtained using the method of Ordinary Least Squares. The numerical values of coefficients of the independent variables can be used to substitute the respective values in Eq. (3); the level of significance of each variable is shown in parentheses. TFPG = −0:091 +

0:017 EXPGROW 0:002CCAPIN 0:420GRO + − ð0:000Þ ð0:009Þ ð0:000Þ

+

0:099 CPC1 0:684 CCR4 0:0002 CFSIZE + + ð0:034Þ ð0:010Þ ð0:104Þ

+

0:031 CPC3 0:099 CPC2 + Σα8 + j INDj + ð0:473Þ ð0:114Þ

Table 4 Coefficients. Variable

(Constant) Growth Rate of Output (GRO) Change in Capital Intensity (CCAPIN) Export Growth (EXPGROW) Change in Firm Size (CFSIZE) Change in Industry Concentration (CCR4) Change in Principal Component One (CPC1) Change in Principal Component Two (CPC2) Change in Principal Component Three (CPC3) Adjusted R square Std. error of the estimate Durbin–Watson Dependent variable: TFPG.

INDj included

INDj excluded

B

B

Sig. (p-value)

Sig. (p-value)

−.091 .420

.118 .000

−.067 .450

.015 .000

− 0.002

.009

− 0.002

.009

.017 0.0002 .684

.000 .104 .010

.019 0.0002 .770

.000 .113 .007

.099

.034

.098

.036

.099

.114

.116

.058

−.031

.473

−.031

.465

0.518 0.375 2.118

0.502 0.381 2.086

A positive and significant relationship between total factor productivity growth and output growth can be observed, thereby confirming that Verdoon's Law holds for the industry sample tested in this study. The output growth which is measured as the sales growth over a year is a highly important variable. The equation indicates that a unit change in output growth causes an incremental change of 0.42 in the total factor productivity growth, ceteris paribus. The ‘change in capital intensity’ too has a positive relationship with total factor productivity growth. The equation indicates that a unit change in ‘change in capital intensity’ causes an incremental change of 0.002 in the total factor productivity growth, ceteris paribus. The growth of exports yielded the expected positive sign. The equation indicates that a unit change in export growth causes an incremental change of 0.017 in the total factor productivity growth, ceteris paribus. The variable ‘change in firm size’ has a very marginal impact on the total factor productivity growth. The equation indicates that a unit change in ‘change in firm size’ causes an incremental change of 0.0002 in the total factor productivity growth, ceteris paribus. In the case of the ‘change in concentration ratio’, an increase of the ratio appears to exert a pro-competitive effect on total factor productivity growth. The equation indicates that a unit change in concentration ratio causes an incremental change of 0.684 in the total factor productivity growth, ceteris paribus. The variable CPC1 (the change in principal component one) shows a reasonably high positive correlation with total factor productivity growth. This principal component represents the cluster of technologies namely, Local Area Networks, Computer Aided Design, Programmable Logic Controllers and Computer Numerical Control Machine Tools. These technologies are used in the manufacturing set-up as ‘process control’ technologies. The corollary is that a unit change in investments in process control technologies causes an incremental change of 0.099 in the total factor productivity growth, ceteris paribus. The variable CPC2 (the change in principal component two) shows a reasonably high degree of positive correlation with total factor productivity growth. This principal component represents the cluster of technologies namely, Automated Storage and Retrieval Systems, Automated Inspections and Robotics. These technologies are used in the manufacturing set-up as ‘production and quality control’ technologies. The corollary is that a unit change in investments process control technologies causes an incremental change of 0.099 in the total factor productivity growth, ceteris paribus. Although the variable CPC3 (the change in principal component three) shows a negative correlation with total factor productivity growth, it can be excluded from the computation as the relationship is very highly insignificant. 4.3. INDj excluded According to the output of the model that excluded the INDj, the significance of explanatory variables, GRO, EXPGRO, CCAPIN, CCR4 CFSIZE, CPC1, CPC2 and CPC3 are, 0.000, 0.000, 0.009, 0.007, 0.113, 0.036, 0.050, and 0.465 respectively. The only significant deviant


observation is CPC2 which was marginally significant in the former TFPG model is significant in this model. Hence, it is inferred that the reliability of TFPG model which included INDj variable is higher, though of course marginally. The variables, CPC1 and CPC2 that denote the changes of the first two PCs (PC1 and PC2) represent two important themes (dimensions), namely ‘process control’ technologies and ‘production and quality control’ technologies that together account for 53% of the variance, which is significant. CPC3 that denotes change in PC3 which gives the dimension of ‘general control’ technology is very highly insignificant and it only accounts for 13% of the variance in FMT. According to both TFPG models, the null hypotheses that CPC1 and CPC2 have no partial correlation with TFPG (i.e. α6 = 0, and α7 = 0) can be rejected. Therefore, the alternative hypotheses can be accepted which means that FMT has a significant correlation with TFPG which is positive (since in both models α6 and α7 are positive). This leads to the acceptance of the research Hypothesis: A high degree of FMT adoption enhances TFPG of the manufacturing industry of Malaysia. Since outliers may exert greater influence on the least squares estimates of the regression parameters than do other observations, the robustness of the model and sensitivity of the model to the outliers were tested as explained in Appendix A. Further, it was revealed that the potential for any particular three-digit industry having a significant impact on the output of the model is marginal (see Appendix B). 5. Conclusions and policy implications The objective of this paper was to evaluate the influence of the degree of adoption of Flexible Manufacturing Technology on the productivity of the manufacturing industry of Malaysia. The types of Flexible Manufacturing Technology considered are namely, Computer Numerical Control Machine Tools, Numerical Controlled Machine Tools, Robotics, Programmable Logic Controllers, Automated Inspections, Automated Storage and Retrieval Systems, Computer Aided Design and Local Area Networks. The presence of multicollinearity among the eight types of Flexible Manufacturing Technology necessitated the use of three PCs to substitute the individual Flexible Manufacturing Technology variables. The Flexible Manufacturing Technology variables load onto PCs as follows: Local Area Networks, Computer Aided Design, Programmable Logic Controllers and Computer Numerical Control Machine Tools load onto PC1; Automated Storage and Retrieval Systems, Automated Inspections and Robotics load onto PC2 and Numerical Controlled Machine Tools only loads onto PC3. The three Principal Components (PCs) were labelled so that they best describe their respective constituents; PC1-‘process control’ technologies, PC2-‘production and quality control’ technologies and the PC3-‘general control’ technology. In this study, two separate models for total factor productivity growth were solved for two situations: one included the Industry Fixed Effects Dummy Variables and the other excluded it. One of the important contributions of the present study is that it reveals, regarding the models specified to study the impacts of Flexible Manufacturing Technology, that by including an industry fixed dummy variable to account for the differences in technological opportunity among different industries, the credibility of the models can be increased at least marginally. The most significant finding of the study is that the change in PC1 shows a significant and positive correlation with total factor productivity growth whereas the change in PC2 shows a marginally significant and positive relationship with total factor productivity growth. This indicates that the increase in process control technologies and production and quality control technologies have direct influences on total factor productivity growth of the Flexible Manufacturing Technology intensively adopted sub sector of the manufacturing industry. In contrast, the change in PC3 shows a highly insignificant

401

and negative relationship with total factor productivity growth. Since both PC1 and PC2 together account for (53%) greater variation and PC3 account for (12%) relatively smaller variation among the eight Flexible Manufacturing Technology, it can be concluded that a high degree of Flexible Manufacturing Technology adoption enhances total factor productivity growth of the Manufacturing Industry of Malaysia. This is in harmony with the a priori expectations regarding Flexible Manufacturing Technology but contrary to the findings of the studies by Brendt and Morrsison (1995) and Amato and Amato (2000). However, these studies are different from the present study due to reasons such as differences in technologies considered, non consideration of Industry Fixed Effects Dummy Variables, differences in countries considered and the differences in the explanatory variables considered. In this regard, the manufacturing sector has been the engine of economic growth since structural transformation took place in the Malaysian economy in 1987. The sustainability of higher economic growth continued to be driven by productivity through the enhancement of TFP. In this regard, TFP development strategies emphasised on the quality of the workforce, raw material, capital structure and technical progress. However, the instability of TFP contribution to manufacturing sector industries in terms of average annual growth rates are dependent on the inputs used in the production which were reported to be insufficient and of low quality. The starting point for policy recommendations is to offer policies that can help to overcome the following main problems of the manufacturing sector, especially the efficiency and productivity being input-driven rather than TFP productivity-driven. Meanwhile, for any industry to develop there must be a regular and consistent supply of raw materials. One of the main problems faced by the Malaysia's manufacturing sector industries is in the supply of raw materials. The manufacturing sector is dependent on imported raw materials, which form the largest component of cost in the Malaysian manufacturing sector. This can have serious adverse impact on the Malaysian Balance of Payments as shown in the Annual Report of Bank Negara (1991–2005) which reported that imported raw materials constituted 20% of the raw materials utilised by resource-based industries while non-resource-based industries as much as 60% of the required raw materials. In particular, leading industries in the manufacturing sector such as electronics and electrical machinery can have imported raw materials content as high as 70% of the total cost. For food manufacturing industries, the sources of supply of raw materials are 70% dependent on imported raw materials. The supply of raw material is not consistent and of low quality in most manufacturing sector industries in general and food manufacturing industries in particular. Besides, shortage of skilled labour also causes a serious constraint on capital utilisation. Skilled labour is required to operate the new technologies embodied in new plants and equipment so that available capital stock may be utilised efficiently. Hence, skills training and the deepening of skills are of vital importance for the full utilisation of capital. Improvement of the quality of the local raw materials will help to improve the final product and enable it to compete in the international markets; and also help to reduce the dependency of the manufacturing sector industries on imported raw materials. These, if attained, could help the industries to become efficient, dynamic, and internationally competitive. Moreover, one of the major problems of Malaysia's manufacturing sector is that the sector is highly dependent on foreign direct investment FDI. The local small and medium scale industries have financial problems compared with the large-scale industries. Getting capital at the right time will save the production of these SMI's. Overcoming the financial problems of industries will improve the productivity of the sector. In addition, low technology has been identified as a major constraint facing the local small-scale industries (SMI's). The findings of

402


ogy investments such as FMT to customer values (Sade et al., 2009) for example, is becoming more crucial as manufacturing concept of productivity moves toward the concept of customerisation. Further, it can be safely admitted that the accuracy of findings can be increased by considering investments in FMT rather than the degree of adoption of FMT. Hence, it is proposed that future studies need be undertaken in collaboration with the industry monitoring institutes of the state sector that makes establishments obligatory to divulge investments made in FMT to evaluate the impact of investments in FMT on total factor productivity growth.

this study are in line with the above-mentioned statement reflecting the relationship between the technological inputs and the scale of production of SMI's. Low technologies are adopted in the manufacturing processes, and manual handling of materials is applied, with low quality control. The first step for improving the productivity growth and efficiency of the manufacturing industries will be to modernise the technology used by small-scale industries to improve the quality of the manufacturing products, and change their production methods. This must be started right from the cultivation of agricultural raw material in order to reduce the harvesting loss, and also to get good raw material quality. The local large-scale industries, on the other hand, are dependent largely on imported technology. For a more sustainable development of the large-scale industries, this imported technology should be kept to a minimum in the short run, while in the long run efforts are made to produce all the technological inputs locally. This can be achieved by adopting the experience of industrial countries and by capitalising on the benefits of global information and communications technology and research done in this area through collaboration with the developed countries and their companies. Furthermore, the level of skilled labour employed would reflect on the level of technology adopted. Therefore, before any improvement on technological and material inputs, there is a need to reduce the number of unskilled labour that dominated the manufacturing sector, and increase the volume of skilled labour in the sector. Concomitant with technology enhancement and as industries become more capital intensive, the critical shortage of skilled manpower will continue for some time. A programme could be designed to upgrade labour standards and use high technology in production methods, through institutions involved in the area of technology skills training for local workers. Finally, since a direct relationship between the total factor productivity growth and the degree of investment in flexible manufacturing technology has been established, the policy makers can focus more on the promotion of investments in flexible manufacturing technology in formulating incentive schemes for the manufacturing industry. The threat of global competition and the changing demand of customers are forcing the Malaysian manufacturing industry to reevaluate their existing operational and technological capability. The study recommends that managers make the manufacturing processes more efficient by adopting a high degree of flexible manufacturing technology despite their widely discussed operational problems and the unfavourable economic appraisals. FMT allows manufacturers to cope with the ever increasing changing demand of digitally connected 21st century customers. These new kind of customers are in a way becoming closer to the production floor and they expect more customization on the products they are buying. For example, companies such as Dell and BMW offer modular customization to their customers who order through their websites. Perhaps by incorporating these kind of customer values into the equation will help to shed some lights and arrive to a favorable economic appraisal for FMT investment in manufacturing. In its customary call for future research, the authors recommend studies that investigate the relationship of investments in FMT rather than the degree of adoption of FMT have with the total factor productivity growth of the manufacturing industry. Relating technol-

Appendix A In order to evaluate the influence of extreme observations Mahalanobis distance and Cooks distance (which indicate the impact of outliers) were obtained from the SPSS output. Mahalanobis distance measures the influence of a case by examining the distance of cases from the means(s) of the predictor variables(s). The critical value of Chi-square, at α = 0.001 level of significance is 51.179 which was taken as the critical value for the Mahalanobis distance; using this value, 12 multivariate outlying cases were identified. Since the values greater than this are problematic and cause for concern they were scrutinised carefully. Further, although they are extreme observations it is not unexpected in a situation where the sample is large (300 cases) and there are a large number of independent variables. Therefore, the outliers were retained in the data set without exclusion. One statistic that does consider the effect of a single case on the model as a whole is Cook's distance. Cook's distance is a measure of the overall influence of a case on the model and it is suggested that values greater than one, may be cause for concern. Only in two cases this critical value was exceeded. Although the exclusion of the extreme cases in both diagnostic tests brought about some changes in the coefficients, they were not significant. Hence, it can be concluded that extreme observations are not cause for concern in the model. Appendix B It is important to check whether any particular three-digit industry can influence the output of the model in a significant manner. The two three-digit industries namely 292 (manufacture of special purpose machinery) and 321 (manufacture of electronic valves and tubes and other electronic equipment) are different from the others. The 292 three-digit industry contains seven five-digit industries encompassing a total of 712 establishments. These two features make it stand out from rest of the three-digit industries. The 321 three-digit industry is a sector which adopts FMT quite intensively. The regression model was solved twice excluding three-digit industries 292 and 321 separately on each attempt, one industry at a time. Table 5 displays the Adjusted R square and coefficients (B) and significance of the three CPCs for each attempt along with the values obtained for the two occasions when all the industries were included and then excluded. According to the table, the exclusion of the three-digit industries 292 and 321, separately has changed Adjusted R square only marginally. Further, the significance levels of CPC1 have changed only minutely when the three-digit industries 292 and 321 are excluded from the model, separately. A similar situation is observed with CPC2

Table 5 Comparison of outputs. Number of industry

Adjusted R square

CPC1

CPC2

CPC3

B

Sig.

B

Sig.

B

Sig.

All 16 industries included All 16 industries excluded Only industry 292 excluded Only industry 321 excluded

0.518 0.502 0.563 0.521

0.099 0.098 0.083 0.109

0.034 0.036 0.034 0.032

0.099 0.116 0.125 0.114

0.114 0.058 0.087 0.092

− 0.031 − 0.031 0.000 − 0.054

0.473 0.465 0.995 0.290


when the same comparison is made. The reason for the situation is that although the excluded industries are significantly different from the others, the difference is not sufficient enough to make an impact on the output that results from the analysis of model because there are 15 other industries at any given attempt. This provides further evidence to the robustness of the regression model used in the study. References Abdullah, M., Hussein, M.A., 1993. Total factor productivity growth in Malaysian resource based industries. ASEAN Economic Bulletin 10 (1), 83–97. Amato, L.H., Amato, C.H., 2000. The impact of high tech production techniques on productivity and profitability in selected U.S. manufacturing industries. Review of Industrial Organization 16 (4), 327–342 June. Bank Negara Malaysia Annual Report Various Issues. (1991–2005) Government of Malaysia printers, Kuala Lumpur, Malaysia. Bloch, H., Tang, S.H.K., 1996. Technical change and total factor productivity growth: a study of Singapore's manufacturing industries. Applied Economics Letters 1999 (6), 697–701. Berndt, E.R., Morrsison, C.J., 1995. High tech capital formation and economic performance in USA manufacturing industries: An exploratory analysis. Journal of Econometrics 65, 9–43. Chandrasiri, 2003. Technological Issues in Small and Medium Enterprises in Sri Lanka. Sri Lanka Economic Journal 4 (1). Chau, K.W., 1993. Estimating industry-level productivity trends in the building from building cost and price data. Construction Management and Economics 11, 370–383. Choong, L., Tham, S.Y., 1995. Total factor productivity in the Malaysian manufacturing sector preliminary results. Jurnal Ekonomi Malaysia 29, 9–35. Coakes, S.J., Steed, L., Price, J., 2008. SPSS 15.0, Analysis without Anguish. John Wiley & Sons Australia, Ltd. Da Silveria, G., Fogliatto, F.S., 2005. Effects of technology adoption on mass customisation ability of broad and narrow market firms. Gestao & Producao 12 (3), 347–359. Denison, E.F., 1967. Why growth rates differ. Brookings Institutes, Washington DC. Department of Statistics Malaysia (DOS), 2000–2005. Annual Survey of Manufacturing Industries — 2000–2005. DOS, Malaysia. Elsadig, M.A., 2006a. Is Malaysia's manufacturing productivity growth input driven? Journal of American Academy of Business, Cambridge 9 (2), 223–229. Elsadig, M.A., 2006b. Carbon dioxide emissions, impact on Malaysia's manufacturing productivity growth. World Review of Science, Technology and Sustainable Development 3 (1), 58–69. Elsadig, M.A., 2006c. ICT and Human Capital Role In Achieving Knowledge-based Economy: Applications On Malaysia's Manufacturing. Journal of Information and Knowledge Management 5 (2), 117–128. Elsadig, M.A., 2007. Biochemical oxygen demand emissions impact on malaysia's manufacturing productivity growth. Global Economic Review 36 (4), 305–319. Elsadig, M.A., 2008a. Impact of foreign direct investment on Malaysia's productivity growth. Advances in Management 1 (6), 28–33. Elsadig, M.A., 2008b. Green productivity; applications in Malaysia's manufacturing. Universiti Putra Malaysia Press, Selangor, Malaysia. Fatimah, S., Mohd, S., 2004. Total factor productivity growth in the Malaysian manufacturing sector. IIUM Journal of Economics and Management 12 (2), 131–163. Federation of Malaysian Manufacturers (FMM), 2007. Directory -2007. FMM. Field, A., 2005. Discovering statistics using SPSS. SAGE Publications Ltd, UK.

403

Fine, C.H., Freund, R.M., 1990. Optimal investment in product–flexible manufacturing. Management Science 36 (4), 449–466 (April). Gale, H.F., Wojan, T.R., Olmsted, J.C., 2002. Skills, flexible manufacturing technology, and work organization. Industrial Relations 41 (1), 48–79 January. Goldar, B., Kumari, A., 2003. Import liberalisation and productivity growth in indian manufacturing industries in the 1990s. The Developing Economies, XL1-4, pp. 436–460. December. Henricsson, P., Ericsson, S., 2005. Measuring construction industry competitiveness: a holistic approach. Research Week International Conference, The Queensland. Jorgenson, D.W., Gollop, F., Fraumeni, B., 1987. Productivity and US economic growth. Harvard University Press, Cambridge, MA. Leung, H.M., 1997. Total factor productivity growth in Singapore's manufacturing industries. Applied economics Letters 1997 (4), 525–528. Mahadevan, R., 2002. A frontier approach to measuring total factor productivity growth in Singapore's services sector. Journal of Economic Studies 29 (1), 48–58. Malaysian Industrial Development Authority (MIDA), 2007. Malaysia — investment in the manufacturing sector. MIDA. McGuckin, R.H., Streitwieser, M.L., 1996. The effect of technology use on productivity growth. Centre of Economic Studies (CES). CES 96-2 April. Menon, J., 1998. Total factor productivity growth in foreign and domestic firms in Malaysian manufacturing. Journal of Asian Economics 9 (2), 251–280 Summer. Oulton, N., O' Mahony, M., 1994. Productivity and growth: a study of British industry, 1954–1986. Cambridge University Press, Cambridge. Porter, M.E., 1990. The competitive advantage of nations. Macmillan, London. Rao, P.S., Preston, R.S., 1984. Inter-factor substitution, economic of scale, and technical change: evidence from Canadian industries. Empirical Economics 9 (4), 247–262. Roller, L.H., Tombak, M.M., 1993. Competition and investment in flexible technologies. Management Science 39 (1), 107–114 January. Sade, A.B., Yusoff, W.F., Shualdi, M.K., 2009. Exploring Web-based Services of Value to a Potential Niche Segment of Biotechnology Researchers in Malaysia. Journal of American Academy of Business, Cambridge, Vol. 15, September 2009. Sharma, O.P., 2002. Managing flexibility in manufacturing and operations. Journal of Management Research 2 Number 3, 147–163 December. Sharma, K., Jayasuriya, S., Oczkowski, E., 2000. Liberalisation and productivity Growth: The Case of manufacturing Industry in Nepal. Oxford Development Studies 28 (2). Shiraz, J.K., 1973. Market structure and price cost margins in United Kingdom manufacturing industries, pp. 67–76. Slagmulder, R., Bruggeman, W., 1992. Investment justification of flexible manufacturing technologies: inferences from field research. International Journal of Operations & Production Management 12 (7/8), 168–186. Solow, R.M., 1956. The production function and the theory of capital. Review of Economics Studies XXIII, 101–108. Solow, R.M., 1957. Technical change and the aggregate production function. Review of Economics and Statistics 39, 312–320. Tham, S.Y., 1997. Determinants of productivity in the Malaysian manufacturing Sector. ASEAN Economic Bulletin 13 (3), 333–343. Wind, J., Rangaswamy, A., 2001. Customisation: The Next Revolution in Mass Customization. Journal of Interactive Marketing 15 (1), 13–31 Winter. Wysokinska, Z. (2004) Competitiveness and its relationships with productivity and sustainable development, Fibres & Textiles in Eastern Europe. Yean, T.S., 1997. Determinants of Productivity Growth in the Malaysian Manufacturing Sector. ASEAN Economic Bulletin 13 (3). Zhi, M, Hua, G.B., Wang, S.Q., Ofori, G., 2003. Total factor productivity growth accounting in the construction industry of Singapore. Construction Management and Economics 21, 707–718 October.