Efficiency and innovation in selected Malaysian government-linked companies for the period 2003 to 2008

The purpose of this study is to measure and assess the performance of selected Malaysian government linked companies (GLCs). A modified strictly output-oriented data envelopment analysis model is used to measure the relative performance of each company by utilizing a list of normalized performance indicators for the period 2003 to 2008. The data envelopment analysis (DEA) scores indicate that ten and twenty-one of the companies are operating on the best-practice frontier under the assumptions of constant return to scale and variable return to scale respectively. Some of the relatively large companies show serious scale inefficiency and exhibit decreasing return to scale. The Malmquist index of total factor productivity is used to investigate the extent of technical efficiency change and technological progress experienced during the period under study. It was found that most companies experienced technical efficiency change but not technological progress (adaptation of technology rather than adoption of new technology or innovation).


INTRODUCTION
The efficiency and performance of government companies (or government-linked companies, GLCs) can be categorized into two views, viz. the managerial view and political or ownership view.The managerial view argues that it is the competition rather than ownership that is the key to efficiency.Competition plays important role not only to achieve efficient allocation of resources, but also as a source of discipline to reduce managerial slack.Studies by Caves and Christensen (1980), Martin and Parker (1995) and Kole and Mulherin (1997) revealed the importance of competition in product market in determining firm efficiency.Given sufficient competitive conditions and no discriminatory regulations and subsidies, both the government and private companies are equally efficient (Borcherding et al., 1982;Millward, 1982;Caves and Christensen, 1980).Hence, to improve *Corresponding author.E-mail: nordinhm@um.edu.my.
efficiency and performance, GLCs should be run on a commercial basis and should compete, not only among each other, but also with other private companies.
The political or ownership view, on the other hand, argues that government companies are inherently less efficient due to various other objectives imposes on GLCs, which are in conflict with the profit maximization objective.Profitability is seldom a top priority in government companies (Vickers and Yarrow, 1991;Boycko et al., 1996).GLCs have to forgo maximum profit in the pursuit of social and political objectives, such as high wage and employment, equity participation and wealth redistribution.Besides distorting the profit maximization objective, the political interference would impose constraints face by managers, such as the pressure to hire politically connected people rather than the best qualified managers and to employ excess labour inputs.Poor efficiency performance of government companies also to some extend due to lack of monitoring system to monitor managerial behaviour, leaving them considerable discretion to pursue personal agendas.Government subsidies, loan guarantees and trade protection removes the risk of bankruptcy and hence lowers managers' incentive to improve performance.
GLCs in Malaysia are defined as companies in which the government has the controlling stake (Malaysia, 2007).The controlling stake refers to, among others, the government's right to appoint members of the board of directors, nominate senior management staff, award of tenders and contracts and provide policy direction for GLCs either directly or through government-linked investment companies (GLICs).While GLCs undertake business activities, GLICs on the other hand, act as an investment arm of the government, investing in selected and strategic companies, including investment in GLCs.
In December 2007, thirty-nine entities were categorized as GLCs and five as GLICs.These 39 GLCs were listed on Bursa Malaysia, accounting for 4.17% of the total number of listed companies and 61.77% of market capitalization of Bursa Malaysia.Five entities identified as GLICs were the Employees Provident Fund, Khazanah Nasional Berhad, Lembaga Tabung Angkatan Tentera, Lembaga Tabung Haji and Permodalan Nasional Berhad.
GLCs have to assume greater roles in Malaysia's economic development.Apart from involving in business activities, they have to undertake other socio-economic objectives.These include, among others, the responsibility to enhance Bumiputeras participation in business activities, employment creation, human capital development and poverty eradication.However, over the years, especially after the 1997/1998 Asian financial crisis, GLCs seemed to be underperformed.Since the last 15 years, GLCs have been laggards in almost all key financial and operational indicators.A study on 15 largest GLCs revealed that only seven companies earned economic profits in 2004 (Malaysia, 2007).
In an effort to improve the performance of GLCs, in May 2004, the Malaysian government decided to reform the GLCs into high performing organizations.Government released the GLC transformation manual containing five policy thrusts that GLCs must adhere to.These policy thrusts clarify the role of GLCs in the context of Malaysia development, outline various measures to improve the effectiveness of the board of directors, enhance GLICs capabilities as professional shareholders, adopting corporate best practices and implementing the GLCs transformation programmes.
The GLCs transformation programmes cover four phases over the periods of 10 years, beginning in May 2004 until 2015.The first phase (May 2004to July 2005) focused on mobilisation of resources, diagnosis and planning of specific programmes for GLCs.The second phase (August 2005 to December 2006) concentrated on generating momentum for GLCs to implement fully the policy guidelines and initiatives identified in Phase 1 and early sustainable improvements are expected to be realised within this stage.Tangible and sustained benefits on the short and medium run are expected to be reaped over the third phase (January, 2007to December, 2009).The full national benefit is expected to be realised in the fourth phase (2010 until 2015).
Given the pressures on GLCs to enhance their efficiency through radical reform program, hence it is crucial to assess the improvement in GLCs economic efficiency, viz.both the technical, technological and scale efficiency.Numerous studies have investigated the performance of GLCs by using the financial ratios, such as returns on sales, assets and equity.These financial ratios examine only a part of firm's business activity (as represented by numerator and denominator of the ratios) and hence provide only partial assessment of firm performance which is generally inadequate to appraise the overall performance of GLCs.
Data envelopment analysis (DEA) technique, on the other hand, calculates the performance indicators that encompass multiple inputs and outputs, provides sufficient information on causes of inefficiencies and economies of scale.It is an effective method in evaluating the overall performance.Hence, the objective of this study is to empirically assess the performance of Malaysian GLCs by utilising the DEA approach.

LITERATURE REVIEW
Numerous studies have been carried out to investigate the performance of state-owned or GLCs and produced varying results.Studies by Megginson et al. (2004), Wei et al. (2003), Dewenter and Malatesta (2001), Ahuja and Majumdar (1998), Kumar (1993), Haririan (1989), Ayud andHegstad (1986), andRamanadham (1984) have shown that government companies tend to be inefficient and ineffective.By conducting a cross-sectional comparison of the conventional accounting ratios (return on sales, return on assets and return on equity) during the period 1975, 1985and 1995, Dewenter and Malatesta (2001) found that government-owned firms are significantly less profitable than privately owned firms.Their findings provide the evidence to support earlier findings of Boardman and Vining (1989) which found that government firms were less profitable and less efficient than private firms.
In a study of 68 Indian state-owned manufacturing enterprises over a period of 1987 to 1991, by using DEA technique, Ahuja and Majumdar (1998) found that stateowned enterprises were operating with low level of efficiency in resource utilization on average of less than 0.35.By comparing the financial performance of stateowned, private-owned and mixed state-private owned enterprises in India during 1973to 1989, Kumar (2003) found poor performance of state-owned enterprises.Similar findings of poor performance of state-owned enterprises in India were also found in earlier studies by Shleifer and Vishny (1997) and Shleifer (1998).Funkhouser and MacAvoy (1979) examined the performance of more than 100 Indonesian public and private companies in the early 1970s and found that public companies were operating at higher costs and earned lower profit margin than private companies.
Studies which examined the performances of GLCs in Singapore included the work done by Ramirez and Tan (2003) and Ang and Ding (2006).GLCs in Singapore are run on commercial basis with no state intervention, receiving no special privileges or subsidies and are expected to be operating efficiently and profitable.In a study of 17 GLCs and 92 non-GLCs for the period 1994 to 1998, Ramirez and Tan (2003) found that in term of access to financing, both the GLCs and non-GLCs in Singapore were competing on equal footing.However, they also found that being GLCs, these companies were rewarded in financial markets with a premium of about 20 percent, reflecting market's perception of benefits (real or illusion) of being linked to the government.Ang and Ding (2006) compared the financial and market performances of Singaporean GLCs and non-GLCs over the period of 1990 to 2000.They found that Singaporean GLCs on average provide superior returns (on assets and equity) and exhibit higher valuations and better corporate governance as compared to non-GLCs.
Despite the importance of GLCs in Malaysia, relatively few studies on the efficiency and productivity measurement have been undertaken.Some of the studies related to the performance of GLCs in Malaysia included the work done by, among others, Hisyam et al. (2008), Issham (2006), Lee (1994) and Nor Idzma and Hetty (2006).Hisyam et al. (2008) compared the financial (proxied by return on assets) and market performances (measured by Tobin Q) of 27 GLCs and 27 non-GLCs over the period 1995 to 2005.Their results suggested that non-GLCs performed better than GLCs on marketbased performance.However, in terms of financial performance, GLCs tend to out-perform the non-GLCs.They concluded that GLCs seems to implement better governance mechanisms and strong management of expenses as compared to non-GLCs.
By utilising the economic value added (EVA) tool, Issham (2006) examined differences in performance of 37 GLCs and 208 non-GLCs over the period 1999 to 2002.His results revealed that GLCs in Malaysia tend to have lower EVA values than the non-GLCs.Issham concluded that his finding is contradicting to the view that government companies have better corporate governance and well monitored, which could be due to the lack of competition effects.When examining the impact of size on EVA values, he found that increase in size of GLCs destroy the company values.It indicates that an expansion of size of GLCs tends to increase the cost of capital in greater proportion than the returns it generates.Lee (1994) examined the aftermarket share price performance of 10 initial public offerings (IPO) of GLCs

DEA METHODOLOGY
The DEA model strictly utilize output-oriented (Mohamad and Said, 2010a) with zero input slacks: maximize Ω0 (1) subject to: (2) (3) Ω0 unconstrained, where Xki and Ykj are observed values of inputs and outputs, i=1,2,..,n and j=1,2,..,m, for each of k=1,2,..,S decision making units (DMUs) and the X0i and Y0j represent the input and output for DMU0 to be evaluated.The efficiency score E0 is given by: .( 4) If (i) Ω0 = 1, and (ii) all slacks are zero, then DMU0 is said to attain full or strong efficiency, that is Pareto Koopman's efficiency.
Otherwise, weak efficiency is attained if only condition (i) is satisfied.For an inefficient, DMU0 say, improvement or movement towards efficient frontier can be identified by inspecting the system of equations with slacks tj + , for all j such that: . ( 5) The projected output is dictated by its peers (identified from λk ≠ 0, for all k) and given by: which can be achieved by proportional increase of (Ω0-1) in all outputs plus additional amount (termed as slacks movements) of tj + in output Y0j whenever tj + ≠0.Thus, (Ω0-1)Y0j + tj + is a measure of underachievement of output Y0j experienced by DMU0.
Models 1 to 3 is the output-oriented model under constant returns to scale, CRS.For evaluation under the assumptions of variables return to scale, VRS an additional convexity constraint is imposed on λk such that: (7) This results in the formation of a convex hull of intersecting planes which envelope the data points more tightly than the CRS conical hull and thus provides technical efficiency scores which are greater than or equal to those obtained under the assumptions of CRS.The difference in technical efficiency scores under the two assumptions of returns to scale is mainly attributable to scale efficiency.Thus, scale efficiency (SE) can be viewed as the extent to which a DMU can take advantage of return to scale by altering its size towards optimal size identified as the region in which there are CRS in the relationship between outputs and inputs and is computed as:

Malmquist productivity index, MPI
In order to further study changes that occurred in technical efficiency and technological practices in two different time periods t and t+1, we utilized the output-based MPI of total factor productivity, TFP as given by Mohamad and Said (2010b): The data covers the 12 months of business operation for each year.
A list of the thirty-one selected GLCs, their abbreviations and core businesses is summarized in Table 1.Investment is the main activity of three GLCs (Astral, MNRB and NCB) while automobile industry is the core business of Proton and UMW.PETRONAS, which is one of the profit-making GLCs has its activities operated by two of its subsidiaries PETD and PETGas.Other activities include hospital support services (Faber), airport management (MAHB), airline operation (MAS), property development (MRCB), highway operator (PLUS), mail services (PMB), Islamic insurance (Takaful), building and construction products (UAC) and the single monopoly of electricity generation, transmission and distribution (TNB).

Selections of input-output indicators
Three input and three output indicators are chosen to characterise and represent the businesses of the GLCs.These indicators are defined as follows: Input 1 (X1): The paid-up capital, expressed in RM'000.Input 2 (X2): The fixed assets, expressed in RM'000.Input 3 (X3): The total salaries and benefits expenses paid, expressed in RM'000.Output 1 (Y1): The sales revenue generated, expressed in RM'000.Output 2 (Y2): The return on assets, expressed in ratio form.Output 3 (Y3): The market price per share, expressed in RM.
The first two input indicators, paid-up capital and fixed assets, Next, we normalize all the variables to a dimensionless parameter with mean unity.This is achieved by dividing all the variables by their respective actual means, which results in observations with values greater (lower) than one to be associated with values above (below) average.A sample of normalized parameters for the year 2008 is depicted in Table 3.Ten companies utilize input X1 (paid-up capital) above the average value, the highest being Maybank (at 3.7360 of the average) while the lowest is Astral (at 0.0344 of the average).TNB is the highest user of input X2 (at 14.105 of the average) while the least user is TEB (at 0.00125 of the average).Sime Darby is the biggest spender on salaries and benefits (at 4.468 of the average) while Lityan spends 0.0148 of the average on salaries and benefits in year 2008.On the output side, Sime Darby generates the biggest sales values of 5.464 of the average, followed by TNB at 4.133 of the average.On return on assets, three companies namely PLUS, TEB and Takaful exhibit high achievements at 7.605, 3.709 and 3.064 of the average respectively.Three most valued businesses as reflected by their market price per share in 2008 were PETGas, Sime Darby and MISC at 2.896, 2.679 and 2.664 of the average, respectively.
Using the normalized data for the period 2003 to 2008, we solve the DEA strictly output-oriented model under the assumptions of CRS and VRS for each year.This amounts to solving 372 DEA models.Results for the mean efficiency scores and returns to scale are summarised in Table 4.

Technical and scale efficiency
Ten GLCs are considered efficient under the assumptions of CRS.These are Duopharma, MNRB, PETD, PETGas, Pharmaniaga, PMB, TNB, TEB, UAC and UMW.These ten companies are also efficient under the assumptions of VRS, thereby implying that they are also 100% scale efficient and are operating on the frontier at the most productive scale size, mpss.Of these ten companies, only two namely TNB and PETGas are relatively capital intensive with paid-up capital of 3.3175 and 1.5145 of the average in 2008 respectively.The other eight are less capital intensive.In fact, Duopharma and UAC recorded paid-up capital of 0.0522 and 0.0570 of the average in 2008 respectively.On the accumulation of asset, we observed that in 2008, TNB recorded the highest asset value of 14.1048 of the average while TEB recorded the lowest asset value of 0.0012 of the average.Yet both are 100% scale efficient and operating at the mpss.
Most of the relatively capital intensive GLCs with paidup capital in 2008 greater than the mean such as the three financial institutions Affin, BCHB and Maybank and the conglomerates such as MISC, Sime Darby and Telekom were found to be operating inefficiently due to serious scale inefficiency.These companes were efficient under the assumptions of VRS.The average PTE score during 2003 to 2008 was 85.33%, suggesting that if these companies were operating efficiently, they could have produced 14.67% more outputs.However, only about half of the GLCs studied were more than 90.0%scale efficient.

Returns to scale
Scale of operations has been identified as one of the contributors to the inefficiency of the inefficient units.DMUs that do not operate at the mpss cannot be fully efficient.These DMUs might be operating under decreasing returns to scale, drs where changing all inputs by the same proportion results in a smaller proportional change in outputs or increasing returns to scale, irs (with a similar appropriate definition).Sixteen GLCs exhibit drs, suggesting that about half of the sample not only experience serious scale inefficiency but also operate in a region of drs.These are Affin, BIMB, BCHB, CCM, Maybank, MAHB, MBSB, MAS, MRCB, MISC, PLUS, Proton, Sime Darby, Telekom and TIME.Most of these companies are large companies and should scale down their scale of operation if they are to operate on the frontier.The remainder five relatively small companies Astral, Boustead, Faber, Lityan and Takaful exhibit irs.These companies should expand their scale of operation in order to become scale efficient.They constitute the potentials for future investments.The mean scale efficiency score for the period 2003 to 2008 was 82.33%, ranging from a minimum of 25.69% (Lityan) to a maximum of 100%.

Malmquist productivity change
The geometric mean of the Malmquist total factor productivity change and its components is depicted in Table 5.On average, the TFP change for the period indicates a small increase of 4.62% per annum, ranging from -15.06 to 28.58%.This is largely due to technical efficiency change rather than technological change (frontier shift or innovation).Hence, adoption of new technology seems to be a problem facing most GLCs as compared to adaptation of existing technology.In other words, the catching up effect which represents the diffusion of technology is dominant in most GLCs.Only six GLCs, all operating at mpss, show improvements in both components.Eighteen (58.06%)GLCs recorded positive TFP growth, while another thirteen (41.94%)

Technical efficiency change (catching-up effect)
With an average score of 1.0944, the TEC showed

DISCUSSION
From the results, we can highlight a few observations, thus: i.Ten (32.26%)GLCs were 100% scale efficient for all years and operating on the efficient frontier at the most productive scale size.
ii.Twenty-one (67.74%)GLCs were technically inefficient.Sixteen of these, mostly large conglomerates exhibited drs while the remainder five (relatively small companies) exhibited irs. iii.

POLICY IMPLICATION
The results stimulate some interesting policy implications regarding future direction and performance of Malaysian GLCs.The study found that ten GLCs were operating on the efficient frontier, mpss for all the years under consideration.These companies were able to sustain their good performance during the entire period.They formed peers and benchmark for the inefficient units.In other words, the inefficient DMUs should imitate and emulate their performance since they were 100% scale efficient.Sixteen GLCs were found to exhibit drs.This suggests over-utilization of input resources, both labour and capital.It involved majority of large companies such as BCHB, Maybank, MAHB, MAS, MISC, Proton, Sime Darby and Telekom.Thus scaling down their scale of operation seems an appropriate action if they were to be on the efficient frontier.This is in line with GLCs transformation guideline 4.7.4 ....option to right size an organisation could include, but are not limited to voluntary separation scheme (VSS), outsourcing, franchising or reengineering certain key activities to drive down the current cost levels, yet maintain (or enhance) the current level of productivity ... (Malaysia, 2005).Another five GLCs were operating under irs.These are relatively small companies whose input resources were below the average (less than 30% of average value in 2008).This suggests under-utilization of input resources, both in terms of quality and quantity, and provides potential for expansion.Thus these companies strongly merit the GLC transformation initiatives provided by the government.
In terms of adoption and adaptation of technology, most firms were found to be adapting well (mean TEC of 9.44% per annum).Firms were found to be closer to the frontier (positive catching up) but the frontier was not progressing as indicated by negative FS (relatively little or no innovation taking place).Since TFP growth involves both changes in TEC and FS, innovation is equally important and efforts should be given to encourage adoption of new technology.A key element to improving GLC operational performance will be to enhance overall operating excellence, both by adoption and adaptation of new technology.The government is now encouraging Mohamad and Said 10269 the firms to explore the avenue of new technology to be adopted and adapted by the firms.Under guideline 4.7.1....in general, GLCs have not divested as much they have invested and/or developed, resulting in several GLCs with unfocused and poor performing portfolios of businesses and activities.For these underperforming non-core businesses, ... unless the GLC believes that there is a very strong strategic and economic rationale to continue to maintain and develop this business, GLCs should be encouraged to identify options to either improve their position (for example through partnerships, including potentially international partnerships) or exit the business (Malaysia, 2005).Lastly, the Government expects the GLC transformation programmes to be a long-term program where the full benefits are only expected to be reaped over an eight to ten year period to 2015.Nonetheless, it is expected that over the short to medium term, tangible results can be achieved with the proper adoption and implementation of the guidelines and initiatives of the GLC transformation (Malaysia, 2005).

Conclusion
The study utilizes a strictly output-oriented DEA methodology to assess the performance of selected Malaysian GLCs using multi-dimensional performance indicators under the assumptions of CRS and VRS.Three input and three output indicators reflecting the physical performance, accounting performance and market value of business are used for the evaluation.Prior to the evaluation data were transformed to normalised dimensionless parameters with mean unity.This enables us to identify values greater (or less) than average to be associated with values greater (or less) than unity.
Results obtained suggest that ten and nineteen GLCs are found to be relatively efficient under the assumptions of CRS and VRS respectively.Thus, only 32.26% are 100% scale efficient with multiple most productive scale size.Most conglomerates exhibit serious scale inefficiencies and are operating in the region of drs, indicating the needs for operation adjustments.These conglomerates should scale down their scale of operations if they were to be operating on the efficient frontier.Five relatively small companies exhibit irs and provide potentials for expansion.These companies should expand their scale of operation if they were to be operating with hundred per cent scale efficiency.
Next, we estimated the Malmquist TFP index and its decompositions using the output-oriented DEA distance functions.The findings indicate that TFP grew at an average rate of 4.62% per annum despite an encouraging catching up effect or TEC of 9.44% per annum which is attributable to both increase in PTEC and SEC.This is due to a decline in the frontier shift or innovation of -3.24% per annum.Only six GLCs were found to exhibit positive growth in all components.Twenty-two and four GLCs exhibited negative growth in FS and TEC respectively.Technical efficiency change contributed more to TFP growth than technology change during 2003 to 2008.Thus, innovation or adoption rather than catching up effect or adaptation of technology is a problem identified facing the GLCs under study.The productivity and underperforming of GLCs were further highlighted and discussed in relation to Guidelines for GLC Transformation Programmes.
The study is by no means complete.Non-GLCs are important entities and contributors to an economy.Its roles as business competitors and economic players cannot be denied.Thus a study on comparison with non-GLCs should be undertaken to complement this study.This is the focus of our next avenue of research.
Further, like any other performance or efficiency evaluation technique, DEA has several limitations.DEA is non-stochastic and does not capture random noise, thereby may have overestimated the magnitude of inefficiencies.The study also assumes that all GLCs under evaluation are fairly homogenous, utilizing similar set of inputs to produce identical outputs.This can only be achieved if we are evaluating a group of firms operating similar business activities such as banking or financial institutions, hospitals and others.The methodology can be revised, expanded and applied to other public and private organizations.
distance function for DMUk with respect to two different time periods under the assumptions of CRS.In other words, if there exist frontier shift (or technological change) in time t+1, efficiency of conversion of inputs in period (t+1) to outputs in period (t+1) relative to technology period t ≠ .MPI, as given by (8) is thus a geometric mean of the productivity changes between two time periods.A value of Mk > 1 indicates positive TFP growth or gain, Mk < 1 indicates TFP decline or loss, and Mk = 1 indicates stagnation or no change in TFP for DMUk from time period t to t+1.The MPI can be decomposed into technical efficiency change, TEC (or catching-up effect) and technological change resulting from shifts in the production frontier, FS (or innovation) such that: Mk(.)=(TEC)k.(FS)k .(1994) further decomposed TEC (relative to CRS frontier) into pure technical efficiency change, PTEC component (relative to VRS frontier) and a residual scale efficiency change, SEC component which captures changes in the deviation between the VRS and CRS technology.Thus the complete decomposition for DMUk becomes: Mk(.)=(PTEC)k.(SEC)k.(FS)k,k=1,2,...,K.The data utilized for the study are annual time-series data for thirtyone selected Malaysian GLCs covering the period of 2003 to 2008.The year 2003 is chosen as the beginning period of study prior to the introduction of the GLCs transformation programmes in 2004.Eight GLCs (Boustead Properties Berhad, Cement Industries of Malaysia Berhad, Malaysia Investment Development Finance Berhad, Opus Group Bhd., Tabung Haji Plantations Berhad, UEM Builders Berhad, UEM World Berhad and VADS Berhad) are excluded from our sample due to unavailability of full data.The data are gathered from two main sources of data base viz. the Bursa Station and Data Stream are supplemented by data from the companies' financial statements as reported in their annual reports.
Four of the GLCs, namely Affin, BIMB, BCHB and Maybank are financial institutions providing banking services (including Islamic banking) while MBSB focuses solely as a property financier.Three are involved in pharmaceutical and health products (Duopharma, CCM and Pharmaniaga).Information and communication technology (ICT) are the main business activities of four GLCs (Lityan, Telekom, TIME and TEB) with Telekom having the monopoly of fixed line communication.Boustead and the conglomerate Sime Darby are two GLCs that are seriously involved in plantations and heavy industries.

Table 1 .
Selected Malaysian government linked companies (GLCs).with the accounting performance.The third indicator, the market price per share is taken as a measure of the market value of the business.Descriptive statistics for all indicators are displayed in Table 2 for the six-year period of study, 2003 to 2008.The input X3 exhibits significant annual variation as opposed to variables X1 and X2.Salaries and benefits experienced an average increase of about 71% from RM 355.30 million in 2003 to RM 608.15 million in 2008.On the output side, variable Y3 is relatively stable as compared to variable Y1.Average annual sales recorded an increasing trend.It increases from RM 3.52 billion in 2003 to RM 6.23 billion in 2008, an average increase of about 77%.The average return on assets exhibits small fluctuation.It drops from 6.944 in 2003 to 6.668 in 2004, steadily increases thereafter to 10.398 in 2007 and drops again to 8.155 in 2008.
denote business operation costs of materials to generate products or to provide services.This may include office rentals, maintenance costs and others.The third input, salaries and benefits expenses are chosen as a proxy for labour input since not all companies discloses information on the total employment.On the output side, the first indicator captures the physical performance of the company.The second, return on assets is equated with the corporate ability to generate a return on its resources and is associated

Table 4 .
Efficiency and returns to scale.
(*) drs and irs refer to decreasing and increasing returns to scale, rts respectively.