Introduction

Last decade has witnessed the impressive improvement in Vietnam Textile and Garment (T&G) industry. Recovery from great difficulties caused by the collapse of the socialist block accompanied with the loss of CMEA market, the industry has developed with the average growth rate of 10.4% per annum and made contribution of average 4% to the overall GDP during 1990s (GSO, 2000). In 2000, the export turnover reached US$2 billion, becoming the largest earner of foreign exchange in Vietnam (Ham, 2000). In addition, the industry is relatively labor intensive, generating the largest employment among manufacturing industries with labor share of 22.2% (VLSS, 97/98). Hence it has been regarded as important and strategic in solving the burden of unemployment and extracting Vietnam’s international comparative advantage in labor intensive production.

The industry, however, has some inefficiency problems which could threaten its sustainable development. Meanwhile export efficiency is rather low with value added of less than 10% (Lan, 2000), utilization capacity rate of the industry is dismal among manufacturing industries with 75% for textile and 81% for garment sub-sector in 1997 (GSO, 1999). The industry has also revealed its vulnerability and weakness to fierce international competition. It was hit hard by the Asian crisis with share of exports to Asian markets falling by 22% and 11% in 1998, 1999 respectively (IOE, 2000). As outdated equipment and technology account for a major part of the industry, although the textile quality has seen some improvements, it fails to meet the demand for high-quality items, making the industry still heavily reliant on imported ones. Meanwhile the garment sub-sector mostly focuses on simple products due to the absence of appropriate technology and highly skilled labor.

As inefficiency has been found in some parts of the industry, serious concerns have been raised about the sustainability of the current pattern of growth. This results in an objective need of studies on efficiency performance, sources of inefficiency and proper policy recommendations to raise the industry’s efficiency. The standard concept of efficiency, however, has been understood and applied in a rather limited and narrow sense. Most researches have normally used some simple accounting indicators to analyze efficiency, which may hide serious performance problems even leading to misleading conclusions in some cases. Meanwhile technical, allocative and scale efficiencies and their links to the notion of frontier functions have been so far best known as only theoretical concepts, with little or no practical application to manufacturing generally and T&G industry particularly. This lack of attention is surprising because these standard concepts of efficiency are important aspect of production, cost and profit economics. It throws a light on the existing performance environment and assesses whether resources are being used effectively in the best proportion. Therefore, an in-depth efficiency analysis with the application of more sophisticated and well-justified methodology as in this thesis has become necessary and significant to fully evaluate efficiency performance of T&G industry. Through it, a better understanding of efficiency could be gained, sources of inefficiency could be identified so that more reasonable and effective policy implications for sustainable development of the industry could be achieved.

Because basic assumptions underlying the cost and profit frontiers for allocative and scale efficiency analysis are not relevant in reality and detailed information on factor demand and input prices is not available, the thesis’s scope is on only technical efficiency analysis. The thesis intends to answer the following central research question

·        What are major determinants of technical efficiency of Vietnam T&G industry?

The thesis firstly reviews the theoretical issues on technical efficiency and its determinants based on a relevant literature survey. Secondly, it focuses on descriptive analysis of Vietnam T&G industry’s performance and features based on secondary data collected from various sources. Thirdly, quantitative and qualitative analysis of determinants of efficiency will be employed based on a chosen econometric model of SFPF and sample of 96 T&G firms in 1998 obtained from “Vietnam Textile and Clothing Competitiveness Survey” conducted by Institute of Economics. Finally, it will be concluded with main findings to suggest some policy implications for further improvement in the industry’s efficiency.

The thesis includes 4 chapters. Chapter 1 provides an analytical framework for efficiency analysis. Chapter 2 introduces an overview of Vietnam T&G industry’s performance and features. Chapter 3 provides empirical results and analysis of efficiency’s determinants. Finally, chapter 4 is devoted to some main findings and discussions on policy implications and limitations of the thesis as well.

Chapter 1. Theoretical Framework

1.1. Concept of Technical Efficiency

The concept of technical efficiency (TE) was firstly introduced in a seminal paper by Farrell (1957), who gave serious consideration to estimating "frontier" production functions in an effort to narrow the gap between theory and empirical work. This concept was then further refined by Forsund, Lovell and Schmidt (1980). Accordingly, assuming a firm employs N inputs X= (X₁, X₂...X_n)' to produce output Y. Thus one representation of production technology is production function or frontier f(X) which sets the upper limit on real output Y, i.e. Y £ f(X). A firm is said to be technically efficient if real output Y coincides with maximum possible output f (X): Y = f(X) and vice versa. Technical inefficiency stems from excessive input usage in production or ineffective resource use, resulting in the condition of cost minimization and profit maximization being not satisfied. In other words, technical efficiency is the ability of the firm to maximize output from a given combination of inputs and technology.

1.2. Measuring Technical Efficiency.

Non-Stochastic Non-Parametric Frontier Models

Farrel (1957) has put the first brick for any further discussion on frontiers and efficiency measurement. In his model, the production frontier can be presented by the unit "isoquant" curve. The curve represents various combinations of the two factors, which an efficient firm uses to produce unit output. Any amount above this curve, which a firm stands, is defined as firm’s technical inefficiency.

Non-Stochastic Parametric Frontier Models

Aigner and Chu (1968) have been the first researchers to compute mathematical frontier form. The functional form they choose is a homogenous Cobb-Douglas function with the requirement of all observations being on or below the frontier. The assumption of constant return to scale as in Farrell's model is not imposed here. Their model can be written as follows:

            lnYi  =   ln f(Xi,b) - Ui                                              (1.1)

where Yi is the i^th firm actual production, f(Xi,b) is the maximum possible output or production frontier, b is parameter vector to be estimated and Ui is one-sided error term (Ui>=0). If Ui is zero, the i^th firm is technically efficient. If Ui is more than zero, there is inefficiency in production. Therefore, TE of the i^th firm can be specified as

            TE= Yi / f(Xi,b)  = exp(-Ui)   (0<= TE <=1)

TE can be measured directly from the residual vector derived from frontier estimation by mathematical programming methods. This model has some apparent shortcomings. Firstly, the estimated frontier is extremely sensitive to outliers. Secondly, the mathematical form may be too simple. Thirdly, parameters estimated have no statistical properties and inferences because no statistical assumptions are imposed on the model.

Non-Stochastic Statistic Frontier Models

The derivation of this model uses the previous functional form (1.1), yet adding statistical assumptions on Xi and Ui. It is often assumed that Ui is independently identical distribution (i.i.d) and Xi is independent of Ui. Different distribution assumptions for Ui can lead to different measures of TE. Afriat (1972) assumes two-parameter beta distribution for exp(-Ui) and Maximum Likelihood (ML) technique is chosen to estimate parameters. Meanwhile Greene (1980) highlights the assumption of gamma distribution should be potentially useful.

Instead of ML method, Richmond (1974) employs Corrected Ordinary Least Square (COLS) method to estimate frontiers and TE. COLS procedure, however, has some problems. Firstly, estimates are very sensitive to assumptions made on statistical distribution of Ui. Secondly, even after constant term correction, some residuals may still have the "wrong sign" i.e. these observations lie above the estimated frontier.

Stochastic Frontier Models

The non-stochastic frontier estimation uses only one-sided error U, which defines exactly the maximum possible output given some set of input quantities. Any variation in each firm's performance is only attributed to variation in the firm's inefficiency. This specification ignores real possibility that in addition to inside factors under the firm control (inefficiency), a firm's performance may be affected also by outside factors beyond the firm's control such as bad weather, external shock, “statistical noise" and measurement error. The stochastic approach could modify the non-stochastic one to single out the “true” efficiency from the mix. Stochastic Frontier Production Frontier (SFPF) was independently proposed by Aigner et al (1977) and Meecusen and van den Broek (1977). Accordingly, the model is defined as

            lnYi = ln f(Xi,b) + Vi - Ui and TE = exp(-Ui)                                         (1.2)   

Vi represents symmetric disturbance to capture the random effects of measurement error, other statistical "noise" and exogenous shocks. Ui reflects the fact that each firm's output Y must lie on or below its frontier (ln f(Xi,b).exp(Vi)). Any such, the deviation from the frontier, which we define as technical inefficiency, is the result of influences of factors under the firm's control. Ui is assumed to be distributed independently of Vi and to satisfy Ui ³ 0. Followings are the two latest specifications: Battese and Coelli (1992) and Battese and Coelli (1995). These are the most general, covering almost SFPF models in the literature.

·        Battese and Coelli (1992) model (2-stage estimation)

            lnYit = ln f(Xit,b) + (Vit - Uit)                i = 1,...,N, t = 1,...,T              (1.3)                                   

where Uit = (Uiexp(-h(t-T))) is one-sided non-negative random term called technical inefficiency effect assumed to be i.i.d as truncations at zero of the N(m,s_U²) distribution.  Many empirical studies (e.g. Pitt and Lee, 1981) have estimated stochastic frontier to measure firm’s TE (stage 1), and then regressed computed efficiencies on firm level variables to investigate source of efficiency (stage 2).

·        Battese and Coelli (1995) model (single-stage estimation)

Although being useful for efficiency analysis, the two-stage estimation procedure has been recognized as inconsistent in it’s assumptions concerning the Ui’s distribution. It is unlikely to provide estimates as efficient as those obtained by using single-stage estimation proposed by Battese and Coelli (1995). This estimation only adds the assumption that Uit is independently distributed as truncations at zero of the N(m_it,s_U²) distribution and m_it = Z_itd where          Z_it is a vector of variables which may affect the firm’s TE and d is estimated vector of parameters.

1.3. Firm Level Factors Affecting Technical Efficiency

Firm Size

Arguments on firm size’s impact on TE have not been consistent. You (1995) believes that an expansion of small firms results in more efficient resource allocation. A large number of small firms may constitute a seedbed for young entrepreneurs who are more creative, active and competent to improve firm efficiency. In addition, efficiency may be higher due to exposure to more competition than larger ones.

However, the opposite conclusion is advocated by a variety of researchers. Jovanovic (1982) assumes a competitive industry with a known time path of future output prices. The firm's cost is denoted as mC(Y) where C(Y) is a cost function common to all firms of the industry and m > 0 is a firm level fixed inefficiency parameter. The firm's static profit-maximization problem is solved by the following expression

max P = PY - m^* C(Y)                                                                      (1.4)

  Y                                                                                                                                            

where P is output price and m^* is the firm's expectation of m conditional on the information available to the firm. From (1.4), Jovanovic obtains ¶Y/¶m^* = -C’(Y)/( m^*C’’(Y)). The firm cost is assumed to be convex, i.e. C’’(Y)>0 thus ¶Y/¶m^* is negative. Therefore, firm size, in terms of output Y, is obviously positively related to efficiency.

Yoo (1992) and Baldwin (1992) argue that larger firms have more capacity and opportunity to bear large fixed costs of replacing equipment, installing modern technology or rooting out inefficiency, which holds that efficiency should increase with firm size. In addition, Caves and Barton (1990) point out small firms tend to be price takers. They are likely to include many new entrants who can suffer from factor market imperfections and/or be subject to shakedown and shakeout.

Firm Age.

Jovanovic (1982) concludes the positive relation between firm age and TE. He highlights that firms consider their efficiency level as given and adjusts operation scale accordingly. All firms have the same m^* and choose the same size in the first period. And then, firms will update their m^* after every period on the basis of difference between expected and realized profits. To approach the correct value of m^*, firms need several periods as unpredictable and firm-level shocks may affect realized profits. Over time, the m^*s will converge to the actual values of firms’ inefficiencies. As a result, efficient firms grow and inefficient ones decline. With assumption of selection effects that firms below some threshold level of efficiency exit the industry, mean TE will increase for groups of firms of the same age overtime. Firm age is consequently positively correlated to efficiency.

Pack (1992) shows that handling and practical experiences overtime with the firm’s machinery can move up the learning curve or locate the firm at a more advanced position in the learning curve. Ericson and Pakes (1995) propose that a firm must decide whether to exit, to continue at current efficiency level or to invest for enhancing efficiency in the beginning of each period. Entrants begin their operation with relatively low investment level. Gradually firms whose investments are successful grow and invest more while those whose investments fail to improve efficiency level maintain their current sizes or leave the industry. Consequently efficient firms are generally larger and older than inefficient ones. Alternatively, a negative efficiency-age relation is supported by Schumpeter’s theory of “creative destruction”, which states that young firms have higher motivation for changing efficiency. Page (1980), Pitt and Lee (1981) postulate newer firms will possess capital of more recent vintage embodying more advanced technology.

Geographical Location

If firms in the same industry or relevant industries are allocated next to one another, they can take advantage of other firms’ invention and adjust to changes or innovations more easily. Competition also forces these firms to produce closer to the frontier. In addition, if not excessive, geographical concentration might be beneficial to TE as it concentrates all necessary facilities for production into one place (Haddad, 1993). However if infrastructure is very poor, concentration puts pressure on availability of resources and may crowd out the access to limited facilities. Each firm becomes a rival to others in using the common infrastructure system. It would be more costly to get access to necessary conditions for efficient production and hence efficiency level of firms more geographically concentrated may be harmed.

Ownership Structure.

Foreign share in ownership is usually perceived to relate positively to firm efficiency due to  superior in terms of experience, managerial skill, technology and business knowledge. In addition, many researchers accept the negative impacts of public share in ownership on the firm efficiency. Kirkpatrick (1984) states that since most public firms are monopolies or account for a large share of the market, an improvement in performance may represent the exercise of the firms’ monopoly position in the market rather than an efficient enhancement. The Government influences over the public firms’ choice of production techniques are partly responsible for their lower efficiency record (Gopta, 1982 and Perkins, 1983). However, Haddad (1993) finds out firms with high public share in ownership may exhibit less deviation from the efficient frontier. Accordingly, special supports to sectors of national importance may allow public firms in these sectors to reach a large size (hence taking advantages of scale economics) and obtain more technological capabilities compared to smaller-size new, private firms with lack of access to modern technology and know-how.

Export Orientation

There is a large body of arguments that export orientation and efficiency are likely to be highly positively correlated. Market expansion enables firms to utilize all their available resources, which are underused or unused as firms are restricted to limited domestic market. In addition, inward oriented firms focusing on the domestic market do not suffer fiercer competition and are thus expected to be less technically efficient (Cheng and Tang, 1986). Rodrik (1992) states that without export orientation, domestic entrepreneurs are likely to prefer and afford to have the so-called "quiet life”. The production level is still below the maximum level attainable, simply because no one wants to work harder for efficiency enhancement. Why do they have to work hard if competition presents a little threat. Export orientation may impact entrepreneurs’ choice between working and leisure. It is supposed that increasing efficiency requires constant effort and diligence, which cut into leisure. Inward orientation creates less competition for domestic entrepreneurs whose choice is in favor of leisure. Efficiency of course is on a lower path as overall effort declines. Outward orientation would reverse the process. The opportunity cost of leisure increase and hence more efforts will be made to increase efficiency.

Chapter 2. Overview on Vietnam T&G Industry

2.1. General Performance of Vietnam T&G Industry

The collapse of the Socialist block and the CMEA market in late 1980s caused great difficulties to Vietnam T&G industry. To overcome these difficulties, a series of economic reforms were launched at both macro and micro level. Consequently, T&G industry was recovered from the crisis and entered a new period of development since 1992. The industry has enjoyed a relatively remarkable average growth rate of 10.4% per year and contributed average 4% to the GDP (GSO, 2000).

The garment sub-sector has performed very well with the average growth rate of 21.4 % during 90s. The major reason for such an impressive performance is the opening access to EU quota-regulated foreign markets in early 1990s. Besides, some non-quota markets namely Japan, South Korea, Taiwan and ASEAN have been remarkably exploited since late 1998. Although gaining a high development, the garment sub-sector has indicated the vulnerability and weaknesses under fierce international competition. In addition, the sub-sector has mostly focuses on unsophisticated items. Some complicated items, which require high technology and skilled labors, are still beyond the production capacity of the sub-sector. Thus only a part of EU quota categories could be satisfied. The number of quota items fell to 54 in 1995 and declined further to 29 in 1998 (VIR, 2^nd November, 1998).

The textile sub-sector has shown a relatively inefficient performance. Most of the textile products had either low or negative average growth rates Owing to obsolescence of technologies and equipment, the textile quality fails to meet sufficiently the demand for high quality ones and the competitiveness capacity is very low. As a result, the sub-sector has been losing the domestic market to imported textile products. The ratio of T&G imports to domestic textile output value is up to 197% in 1998 from 94% in 1995. (IOE, 2000).

One explanation for difference in performance of these 2 sub-sectors is technology. In general, the garment sub-sector’s technology has been improved much better than that of textile one with up-gradation of nearly 100% of garment firms. Given forex shortage, it is much easier to modernize technologies for garment firms than textile ones. The average capital of US$500,000 is sufficient for a new garment firm to start operation whereas at least US$10 million is needed for textile firms (VIR, 16^th March, 1997). A majority of equipment in textile sub-sector was made in 1960s. Although a number of out-of-dated technologies were innovated in recent years, the rate of replacement was only 5.2 % (Ham, 2000).

Specifically, some textile firms produce both textile and garment products although the vertical linkage in the T&G industry is rather weak. Mixed production, however, is different from the fully integrated production. The motives for opening additional garment production lines are to take advantage of access to export quota for their garment products, to add value to some textile products or simply to exploit the excessive employment.

2.2. Features of Vietnam T&G Industry.

Employment Size

T&G industry has created largest employment among manufacturing industries with labor share of 24% (92/93) and 22.2% (97/98). Overwhelming labor is partly explained by the relative labor intensity of the industry. Informal sector (self-employment) in T&G industry accounts for a large share of total employment (52.86% for 92/93 and 46% for 97/98) (IOE, 2000). Foreign-invested firms generate the smallest number of jobs. Household enterprises, which dominate the local non-state sector, are very small in size, causing the sector’s average labors to be only 3 and 2.4 per establishment for textile and garment sub-sectors respectively. SOEs are the largest with size being 2.6 times and 2.2 times bigger than the foreign firms in textile and garment sub-sectors. Foreign firms, however, have the highest labor productivity with output per employee being 1.4 times in textile and 1.6 times higher in garment than SOEs.

Geographical Distribution

T&G firms are generally located in 2 poles of the country. The South takes the largest share (50%) of the total industry outputs, followed by the North (40%) (MPI, 1998). The Center represents a very small share mainly because of the absence of adequate infrastructures. The uneven distribution of T&G industry is clearest in foreign direct investment (FDI). The South hosts the overwhelming number of foreign-invested projects (93%) with 98% of registered capital. Uneven distribution of FDI in the industry is mainly due to inadequate infrastructures and the big difference in land price in various parts of the country. Moreover, the South has much better connections with overseas Vietnamese, those have played an important role either in attracting FDI and/or in establishing contacts with brokers.

Ownership Structure

·        Ownership Structure of Textile Sub-sector

Although the share of state sector has been fallen in 90s, it is still rather large at 53% in late 1990s. The private sector developed sharply until 1996 but slowed down then. However, even in the years of high growth rate, its share was only 2.1% in 1996/97. Cooperatives, which once played an important role in the centralized economy, are now almost non-existent with its share being only 1% in 1998 (IOE, 2000). The reason for such a sharp decline is the ending of government assistance, poor economic management and low motivation of managers due to improperly incentive system. Meanwhile household sector's output share fall dramatically from 22% in 1995 to 15% in 1999 mainly because of the expansion of formal sectors. Because the textile sub-sector is import competing, which enjoys preferential treatments provided by the Government including tax reduction, high protection levels, etc., foreign invested sector has enjoyed its increasing importance in late 90s with its output share of 31.8% in 1998, second only to SOEs.

·        Ownership Structure of Garment Sub-sector

The state sector is still important in the sub-sector. The role of local non-state sector is modest due to capital shortage and lack of favorable environment to flourish. Private firms often complain about the complicated registration procedures^[1], difficult access to land, credit and quota. Unfavorable business environment has had its negative impact not only on private firms, but also on cooperatives and household enterprises. Too small scale of non-state firms is also widely perceived as a cause of their poor performance and their disadvantage vis-à-vis SOEs and foreign invested firms. Share of foreign-invested sector although having risen by less than that in textile sub-sector is still fairly significant and no less important than that of state sector. Although foreign invested firms do not have opportunities to access export quotas, they can take advantages of cheap labor for transit trade i.e. re-exporting to their home countries or to other countries including EU and the US.

Market Orientation

·        Market Orientation of Textile Sub-sector

Although many efforts have been made, domestic textile products are rarely to penetrate international markets due to low quality and high price. However, these products only meet a small part of domestic demand. Vietnam now still depends heavily on textile imports, which are used as both inputs for T&G production and final consumer products.

·        Market Orientation of Garment Sub-sector.

It is estimated that the share of domestic garment products in local consumption has substantially risen from about 60% in 1994 to 85% in 1997 (IOE, 2000). Generally, the local demand for garment products is mostly satisfied by garment households or tailors with very low service charge meanwhile demand for ready-made clothes is still modest although has started to rise in recent years. A part of local consumption is met by small share of imported items. The volume of imported garment products has been fallen substantially due to the development of local production.

Meanwhile the informal sector focuses on the domestic market, most of the formal sector sets target on the foreign markets. The industry has successfully shifted from traditional CMEA markets to EU and Asian markets after the loss of CMEA markets. Its export value has increased sharply, reaching nearly US$ 2 billion in 2000. Besides very small share of textile exports such as towels, handkerchiefs, and embroideries, garment exports are overwhelming with 95.2% of total textile and garment exports in 1998 (GSO, 2000).

Although the garment sub-sector has sufficient production capacity and availability of export quota grants, 80% of garment exports are in the form of sub-contract for foreign brokers due to poor international marketing skills. Vietnam firms receive very low value added of no more than 20% of recorded export value. The remaining 80% is captured by foreign importers and brokers.

Chapter 3. Determinants of Technical Efficiency of T&G Industry

3.1. Specification and Data

The SFPF model with translog functional form and single-stage estimation approach proposed by Battese and Coelli (1995) is employed to investigate determinants of TE of T&G industry. Accordingly the frontier technology of T&G industry is defined as

lnYi = b₀ + b₁ln(WAGEi) + b₂ ln(INTERi) + b₃ ln(CAPi) + b₄ ln(WAGEi) x ln(INTERi) + b₅ ln(INTERi) x ln(CAPi) + b₆ ln(CAPi) x ln(WAGEi) + b₇ ln²(WAGEi) + b₈ ln²(INTERi)  + b₉ ln²(CAPi) + GAR + Vi – Ui                             (3.1)

Yi is the i^th firm’s total production value, WAGEi is the total wages, INTERi is the value of intermediate inputs, CAPi is the value of net capital stock and GAR is the dummy variable taking the value of 1 for garment firms and 0 otherwise. Vis are assume to be i.i.d N (0, s_v²). Uis are technical inefficiency effects, which are assumed to be independently distributed, such that Ui is the truncation (at zero) of the normal distribution N (m_i, d_u²) where

m_i = d₀+ d₁ ln(SIZEi) + d₂ ln(AGEi) + d₃ ln(SIZEi) x ln(AGEi) + d₄ LOC1 + d₅ LOC2 + d₆ OWN1 + d₇ OWN2 + d₈ EXPi                                                    (3.2)                                                  

Values of the 20 unknown parameters in these two equations are simultaneously estimated by ML method using the computer program FRONTIER 4.1 designed by Coelli (1994). Data and variables for this model are drawn 96 T&G firms in Vietnam Textile and Clothing Competitiveness Survey conducted by Institute of Economics.

3.2. Determinants of Technical Efficiency

Table 1. Maximum Likelihood Estimates for Parameters

^(*) and ^(**) : significant at 1% and 5%, respectively

Source: Author’s calculation based on the sample and empirical result

According to empirical results, the TE frequency distribution of T&G industry is negatively skewed. The TE differs substantially among firms. They range from 21.18% to 97.94% with the mean efficiency estimated to be 87.79%, suggesting that on average, T&G firms produce 87.79% of the output that could be theoretically produced with the same bundle of inputs by a technically efficient firm. In other words, they are 13.21% inefficiency from the frontier. There is a majority of the sample firms (60.42%) having TE greater than 90%. The sample frequency distribution also indicates that the highest number of firms (35.42%) have TE between 90% and 95%, followed by firms enjoying highest efficiency levels located to the right of 95%. Although there are very high relative frequency of TE above 90%, as much as 22.92% of total sample firms are quite poorer in their efficiency performance with more than 15% technical inefficiency. Hence there appears to be considerable room for affecting improvements in TE of the industry.

Followings are major factors affecting the TE performance of Vietnam T&G industry.

Age-Efficiency Effect

To obtain an indication of the age-efficiency relationship, sample firms are categorized into 5 quartiles. The quartile 4 with mean age of 22.53 years and mean size of 1,201.32 workers is the most technically efficient in production.

Figure 1. Mean Technical Efficiencies by Age Quartiles

Source: Author’s calculation based on the sample and empirical result

TE exhibits “concave curve pattern” between quartiles 1 and 4, implying that the positive age-efficiency relationship seems to be stronger for young than for old firms. The reason is that gains in TE become smaller overtime as experience plays an important but gradually decreasing role in production. These gains theoretically will be completely exhausted.

To investigate the sign of the age-efficiency effect for all observations we need to evaluate the partial derivative of the mean of the inefficiency effects, m, with respect to ln(AGE):

¶m/¶ln(AGE) = -2.11 + 0.29 ln(SIZE)                                                           (3.3)

By setting the derivative in Equation (3.3) to zero and solving for size, we obtain SIZE = 1,445. The age-size values for sample firms are plotted together with the line SIZE = 1,445 in the Figure 2. In this figure, an overwhelming majority of firms are on that side of the line where the marginal effect of age on technical inefficiency is negative, implying the positive age-efficiency relationship. A smaller number of firms are located in the right of the line SIZE=1,445, exhibiting a negative age-efficiency relationship. This is one reason why age quartile 5 with mean size of 1,456.80 and many large firms has lower TE level than the previous quartile with the mean size of less than 1,445.

The above finding can be explained as follows. In the T&G industry when the firm remains not so large in operational scale, the experience is very important. Over years, the production practice with firm’s machinery and equipment and other resources may create learning effects. Hence the older firms get more experience and superior in management and production, which could result in higher level of TE. The role of experience, however, is worth only when the firm operates at reasonable size. As the scale of the firm is very large, the management of production becomes more difficult and complicated. In this case, the advantage of experience does not play because the older firms become stagnant and hesitant to remove old and lagged management method, which is in need to bear larger scale. Under this circumstance, the newer firms tend to be willing to employ new management of production, have high motives for improving TE and possess capital of more recent vintage and hence become more suitable in the large size with higher efficiency performance.

Figure 2. Marginal Effect of Firm Age on Technical Inefficiency Effect

Source: Author’s calculation based on the sample and empirical result

Size-Efficiency Effect

Like the previous case, the sample firms are also categorized into 5 quartiles. The first 4 size quartiles enjoy increasing efficiencies ranging from 84.20% to 92.68%. The efficiency of quartile 5 with mean size of 2585.35 is down by 5.74% compared to the size quartile 4.

Figure 3. Mean Technical Efficiency by Size Quartiles

Source: Author’s calculation based on the sample and empirical result

To take into account the interaction between firm age and firm size, we also compute the partial derivative of the mean of the inefficiency effects m  with respect to ln(SIZE)

             ¶m/¶ln(SIZE) = -0.40 + 0.29 ln(AGE)                                                     (3.4)                           

By setting the expression in Equation (3.4) to be zero and solving for age, we obtain AGE = 4 years. To determine the direction of the marginal effect of size on technical inefficiency, a comparable analysis to that in the subsection above is performed in Figure 4. Almost all of observations are in the right of the straight line where the marginal effect of size on technical inefficiency is positive. This means after 4 years of operation, when the size declines, the firm will enjoy higher TE. The negative size-efficiency relationship could be explained by the possibility that after operating for several years and gaining a specific experience in production, the smaller operational scale can result in more efficient resource allocation and it is suitable for the current limited management capacity of the T&G entrepreneurs. Besides this, smaller firms are facing more competition than larger ones and have willingness to make the best use of their non-abundant resources to survive in the fierce market. Therefore, smaller firms are more technically efficient. By contrary, within 4 years since establishment, new firms mostly with small size have lower TE as it takes very short time for them to gain enough necessary experience in production. However, excepts for the case of entire exit from the industry, newly-established firms become older overtime, achieving more experience and hence the small size will the best choice for improving TE.

Figure 4. Marginal Effect of Firm Size on Technical Inefficiency Effect

Source: Author’s calculation based on the sample and empirical result

Geographical Location-Efficiency Effect

The Table 1 shows North-based firms have mean of technical inefficiency effect of 1.72 higher and 0.19 lower than South-based and Center-based firms, respectively. Mean TE for 3 regions are illustrated in Figure 5. The South-based firms perform efficiency of 8.43% more than the Center-based firms and 5.08% more than the North-based firms. The reason for better performance of South-based firms is that the relatively small Southern region is home to majority of T&G firms. Geographical concentration can enable firms to enjoy the positive externality and learning effects through taking advantages of other firm’s invention. The concentration also generates fiercer competition among close firms, forcing to improve TE. South-based firms can also benefit from using favorable facilities for production namely telecommunication, seaport, airport, road etc. In addition, workers in these firms are well trained and more skillful, most entrepreneurs are rather young, creative, active and willing to apply new management method in production. By contrary, the scattered distribution of firms and very poor infrastructure in the Center result in the lack of motivations for local firms to produce closer to the frontier. Due to very low living standard in the region, Center-based firms are facing the problem with lack of good managers, engineers and technicians and the problem of unskilled and unstable workers.

Figure 5. Mean Technical Efficiency by Geographical Location

Source: Author’s calculation based on the sample and empirical result

Ownership - Efficiency Effect

In the empirical result, the negative sign of d₆ and positive sign of d₇ imply that state-owned firms have the highest TE followed by foreign invested firms. Local non-state sector expresses the lowest efficiency level. Mean TE is illustrated simultaneously in Figure 6.

Figure 6. Mean Technical Efficiency by Ownership Structure.

Source: Author’s calculation based on the sample and empirical result

Mean TE is estimated at 90.46% for the state-owned firms (or 9.54% inefficiency). The inefficiency level is 12.89% for foreign firms and as high as 18.04% for local private enterprises. In other words, SOEs are by 8.5% and 3.35% closer to the production frontier than the local non-state and foreign invested firms, respectively. This observation may have two plausible explanations. On the one hand, T&G market, both domestic and foreign, is highly competitive and therefore SOEs face highly competitive pressures and therefore have to strive hard to be efficient. On the other hand, the favorable business environment is clearly in favor of SOEs. Most of quotas to EU, Canada or Norway is going to SOEs through VINATEX as their powerful representative. SOEs are also in a better position to access foreign clients and rationed foreign exchange. Therefore inputs for SOEs are consistently underestimated relative to those for foreign or private firms, either directly - due to the cost of allocated quota being left out[2] , or indirectly - through the fact that private firms are not able to reach the necessary production scales due to the lack of access to capital. At the same time, outputs of private firms may consistently understate the true quantities due to more limited access to foreign clients (even in non-quota markets) and they may therefore have to export through SOEs. In addition, it should be concluded the cases of under-reporting of private firms to avoid taxes and “price-transfer” (report high input cost imported and low price exported to mother enterprise) of FIEs, which underestimates output of these sectors. Furthermore SOEs are in a better condition to attract scarce skilled workers and technicians, who are willing to accept lower wage rates to receive “prestige” in return. To establish a more reliable ownership-efficiency link, all these distortions should be taken into account and somehow quantified (e.g. by means of shadow price analysis).

In short, the relative efficiency of SOEs vis-à-vis non-state domestic and foreign firms should be interpreted with care. A simulation exercise that estimates firm-specific technical efficiencies under various reform scenarios, where distortions are somehow estimated and partially or totally removed, may provide a more accurate efficiency picture.

Export Orientation-Efficiency Effect

The estimated value of coefficient for EXP is -0.01, meaning that when export share of a firm increases by 1%, the mean of technical inefficiency effect will drop by 0.01. In other words, if the firms expand their sales to international markets, their TE performance is expected to be better. That is because more export-oriented firms seem to face fiercer competition in the world market, enabling them to make the best use of their available resources for market expansion, which could result in higher probability of better technical efficiency. To compute how much export oriented firms’ efficiency is higher than that of inward-oriented ones, we use two alternative definitions of two types of firms. Firms with more than 50% (for the first definition) and 70% (for the second) output exported are classified as export-oriented. Some features on age, size and mean technical efficiency of the two types are reported in Table 2.

On average, the inward-oriented firms are by 0.88% and 1.71% farther from the potential production frontier than export-oriented firms. These figures estimated are rather small maybe because of not large sample. However the positive export orientation-efficiency effect then could be concluded.

Table 2. Export Orientation and Technical Efficiency

            Source: Author’s calculation based on the sample and empirical result

3.4. Comparison With Other Approaches

A Comparison with Two-Stage Estimation

Two-stage estimation of the same SFPF model is conducted for comparative purposes. The first stage involves the estimation of Equation (3.1) by ML method, assuming that Uis are i.i.d as truncation at zero of an N(d_u, s_u²) distribution. The second stage involves the regression of the negative logs of TE predictions from the estimated first stage model upon the 5 firm specific factors. The signs of estimated coefficients are the same as those obtained in the single stage estimation procedure. All coefficients, however, are not significant at 5% or 10% level except d₄, which is viewed as a support for the single stage procedure.

Comparison with Approach Using 70% Threshold Dividing Sub-sectors

Chapter 4. Conclusion and Policy Implications

4.1. Conclusions

The most interesting finding is the effects of firm size, age and their interaction on TE of T&G industry. With the proxy of number of workers as firm size, the thesis has found that the size-efficiency effect is negative for firms older than 4 years and positive for those younger than 4 years. In other words, the marginal effect of size may become positive as firms get older. On the other hand, the correlation between age and efficiency may be negative or positive, depending on firm size. As the size is not so large with less than 1,445 workers, older firms will enjoy higher TE. This effect will be opposite as firms are beyond that threshold. Because of the interaction between firm age and size on efficiency, it is impossible to define exactly the best type of firms which perform the best TE. Nevertheless, two groups of not so large, old firms and large, young firms tend to have higher TE than other groups. It should be noted that this conclusion may be sensitive to other size proxies.

In addition to firm size and age, geographical location may also influence firm efficiency performance. The thesis has investigated that Southern Vietnam seems to be the best place for T&G firms. Due to more pressure of competitiveness, better infrastructure and facilities, more skilled labor and management, South-based firms have the highest mean TE of 91.21%, 8.43% and 5.08% higher than Center-based and North-based firms, respectively. The Center seems not to be the favorable place to set up T&G business currently.