Total factor productivity and the learning mechanism by exporting case study for SMEs in vietnam

Policies and Sustainable Economic Development | Total Factor Productivity and the Learning Mechanism by Exporting: Case study for SMEs in Vietnam NGO HOANG THAO TRANG University of Economics HCMC - trangnht@ueh.edu.vn Abstract By using two-step system GMM method, the study tests the learning mechanism by exporting hypothesis of small and medium-sized enterprises (SMEs) in Vietnam from 2005 to 2013 The results show that export has a positive impact on total factor productivity at firm level Moreover, the study finds reasons for an appearance of learning mechanism by exporting: (i) exporting firms receive technological transfers from developed countries; and (ii) exporting firms have enough capabilities to absorb knowledge generated by exporting The findings suggest that SMEs in Vietnam should invest in R&D as well as machinery and equipment, and actively explore and access to developed countries Moreover, the government should support formal credits, internet services, and business environment for SMEs to improve the capacity to absorb foreign knowledge through exporting Keywords: export; total factor productivity; SMEs; learning by exporting; export destination Introduction The micro economic literature on the relationship between export and productivity at firm level has emerged recently Empirical evidence shows that exporters perform better than non-exporters in the terms of productivity These findings may reflect either self-selection of the best firms in export markets or learning by exporting (Roberts & Tybout, 1997; Clerides et al 1998; Bernard & Wagner, 1997; Clerides et al 1998) Moreover, in testing learning by exporting hypothesis in developing countries the results have been quite mixed and so far inconclusive (Bernard and Jensen, 1999; Clerides et al., 1998; Aw et al., 2000; Graner & Isaksson, 2007; Boermans, 2010) According to the survey by Central Institute for Economic Management (CIEM) of Vietnam in 2013, most Vietnamese enterprises were micro businesses (70%), while small- and medium-sized ones accounted for 24% and 6% respectively Enterprises in the formal and informal sector made up 70% and 30% of total SMEs respectively Ownership was mainly households (over 70%) For the export sector, the results showed that only about 6.27% of the total number of SMEs engaged in exports and a number of exporters to the EU were the largest (13.8%) The fact that a small proportion of firms engaged in exporting was contributing to reduced competitiveness of the economy The primary aim of the study is to test the learning mechanism by exporting of SMEs in Vietnam This topic is an essential theme in the context of Vietnam's international economic integration This paper has the following new features Firstly, the study tests the learning by exporting mechanism and finds out why learning by exporting mechanism happened Secondly, the study uses data from 2005-2013 of CIEM This dataset contains many detailed questions about export activities of SMEs Thirdly, the study uses two-step GMM dynamic panel regression analysis to control the endogenous factors in the model Literature reviews and framework 2.1 Literature and empirical studies There were two main schools of theory to explain why the export business are more efficient: the theory of self- selection mechanism and the theory of learning by exporting mechanism The theory of selfselection mechanism showed that only higher productive enterprises participate in export activities and they have enough capacity of the international environment (Roberts & Tybout, 1997; Clerides et al 1998; Bernard & Wagner, 1997) Meanwhile, the theory of learning by exporting argued that export is an origin of the total factor productivity growth The effectiveness of learning by exporting includes knowledge, technology, and efficiency gained from the exporting process The process of learning takes place through technical assistance from international buyers (Grossman & Helpman, 1991) When businesses engage in export activities, they would be able to absorb knowledge from their export partners It would help these enterprises improve their production capacity (Bernard & Jensen, 1999; Wagner, 2002) International consumers and competitors would transfer knowledge and technology to domestic exporting firms This leads to technology transfer from the traditional technology to modern technology (Rodrik, 1988; Grossman & Helpman, 1991; Clerides et al., 1998) When foreign demand requires a certain level of standards, the importers in the developed countries may provide the technology for exporting firms in developing countries The reason is that the production techniques in developing countries not meet the quality standards of international exporting markets Moreover, the learning by exporting theory argues that exporters in developing countries would gain more benefit through exporting if they export more goods and services to developed countries than to developing countries, This is because developed countries have higher qualified technology, higher technical requirements, and therefore have highly conditional transfer of modern technology for the exporters in less developed countries (De Loecker, 2004; Graner & Isaksson, 2007; Eaton et al., 2008) The majority of empirical studies only found exporting self-selection mechanism rather than learning by exporting mechanism (Bernard & Jensen, 1999; Clerides et al., 1998) However, some recent studies found results of learning mechanism through exporting (Aw et al., 2000; Van Biesebroeck, 2005) In addition, empirical studies also found evidence of the role of export- destination to total factor productivity at firm level (De Loecker, 2004; Graner & Isaksson, 2007; Eaton, 2008) 2.2 Other factors affecting productivity at firm level Other factors which have effects on TFP at firm level include firm characteristics, such as size, age, ownership (Barney, 1991; Wernerfelt, 1984); absorption capacities of the firm, such as the level of technology and quality of human resources (Barney, 1991; Cohen & Levinthal, 1990); entrepreneurship (Schumpeter, 1947; Audretsch et al., 2006); competition in industry (Porter, 1980; Nickell, 1996); and business environment (North, 1991; Li et al., 2000; Acemoglu & Johnson, 2005; Cull & Xu, 2005; Augier, 2012; Harris et al 2011) 2.3 Framework for analysis The framework used for analyzing how export affects total factor productivity at firm level is based on literature and empirical studies of neoclassical growth theory, endogenous theory, trade theory, institution theory, resourced based theory, entrepreneurship theory, and competitive advantage theory Therefore, factors affecting total factor productivity at firm level can be divided into several groups: exporting activities (Group 1); firm characteristics and absorption capacities (Group 2); entrepreneurship characteristics (Group 3); industry characteristics (Group 4); business environment factors (Group 5); and industry- and regionrelated factors (Group 6) Characteristic Capability (age, ownership, size) Export Export Absorb and transfer knowledge from outside (technology, employees; innovation) Technological (Institution and infrastructure) TFP growth Destination Entrepreneurs (education, skill) Business environment Industry, region Figure The analytical framework 3.Overview of SMEs in Vietnam from 2005-2013 The study analyzes some important points of SMEs in the period 20052013 including export activities, the expenditure for investment, and the polices that governments should support for SMEs Export activities Table The proportion of export business over time Year Exporting firm Non exporting firms Total firm 2005 6.43% 93.57% 2,816 2007 5.85% 94.15% 2,633 2009 5.84% 94.16% 2,655 2011 6.04% 93.96% 2,518 2013 6.27% 93.73% 2,550 Source: Author’s calculation from CIEM, DoE and ILLSA survey from 2005-2013 Table Destination of export of SMEs overtime Year China Japan 2005 10.5% 11.6% Asean Other 16.6% Asia 17.7% America Europe n 17.7% 27.1% Russia Other outside 4.4% Asia 18.2% 2007 12.3% 11.7% 13.0% 20.8% 25.3% 32.5% 6.5% 13.6% 2009 15.5% 14.2% 14.2% 16.1% 16.1% 25.8% 8.4% 10.3% 2011 9.9% 8.6% 9.9% 17.1% 8.6% 19.7% 7.9% 5.3% 2013 10.6% 11.3% 13.1% 11.3% 8.8% 13.8% 5.0% 8.8% Mean 11.8% 11.5% 13.3% 16.6% 15.3% 23.8% 6.4% 11.2% Source: Author’s calculation from CIEM, DoE and ILLSA survey from 2005-2013 Table The rate of exporting firms have a technology transfer when exporting Year Technology transfer Non Technology transfer23.4% 2005 76.6% 2007 69.4% 30.6% 2009 67.0% 33.0% 2011 67.1% 32.9% 2013 60.3% 39.7% Source: Author’s calculation from CIEM, DoE and ILLSA survey from 2005-2013 The result shows that the proportion of enterprises engaged in export in 2005 was 6.43 % and it had a downward trend in 2007 and 2009 (only 5.85 % of business involved in exporting); this trend tended to recover in 2011 (6.04 % of firms engaged in exporting) However, the overall rate of export enterprises not engaging in exporting tended to increase overtime From the export market side, the percentage of exporting firms to the Europe market is the highest but it tends to decline over time (27.1% and 13.8% in 2005 compared with 2013) The second largest export markets are other Asian and American countries but they tend to decrease relatively (other Asia countries: 17.7% in 2005 compared to 11.3% in 2013; US: 17.7% in 2005 compared with 8.8% in 2013) The third largest export market are Asean countries and it tends to increase again in recent years (9.9% in 2011 compared to 13.1% in 2013) The next export markets are Japan, China and countries outside the Asia and it tends to increase in recent years (China: 9.9% in 2011 compared to 10.6% in 2013; Japan: 8.6% in 2011 compared to 11.3% in 2013; the other countries outside Asia: 5.3% in 2011 compared with 8.8% in 2013) Moreover, the percentage of exporting firms to Russia is the lowest and only account for 6.4% and it tends to decrease over time (7.9% in 2011 to 5% in 2013) In addition, the research shows that export helps technology transfer; however, a number of businesses suppose that level of exporting that has helped technology transfer tends to decrease over time (76.6 % versus 60.3 % in 2005 and 2013 respectively) Investment Expenditure of SMEs Table The percentage of investment in each item Huma Year Equipme R&D n capit nt 2005 53.4% 1.2% 0.6% Patent s 0.2% Investment Land Building s in other enterprises s 11.3% 22.2% 2.4% Other investme nts 8.9% 100% Total 2007 38.2% 0.2% 0.1% 0.0% 20.3% 34.0% 0.6% 6.4% 100% 2009 22.1% 0.3% 0.2% 0.0% 8.4% 16.0% 3.4% 49.1% 100% 2011 17.1% 0.7% 0.1% 0.0% 6.3% 7.8% 0.6% 67.3% 100% 2013 18.1% 0.2% 0.1% 0.1% 5.5% 9.4% 1.6% 65.3% 100% Source: Author’s calculation from CIEM, DoE and ILLSA survey from 2005-2013 The results show that the proportion of money invested in machinery and equipment decreased over time (53.4 % in 2005 compared with only 18.1% in 2013) The proportion of money invested in R&D activity is very low and tends to decline over time (1.2 % in 2005 compared with 0.2 % only in 2013) Similarly, the proportion of money spent on employee training and patents is very low and almost close to Meanwhile, other investment rates tend to increase sharply (8.9% in 2005 compared to 65.3% in 2013) The importance of an assistance from the government Table Evaluation the important assistance of the government that SMEs need Micro SMEs Removing bureaucratic requirements/restrictions 7.8% Assistance with premises/land 12.7% Small SMEs 13.5% Medium SMEs 14.5% 14.7% 15.9% Assistance with infrastructural facilities 4.6% 3.0% 1.6% By providing easier access to credit 24.8% 28.7% 31.6% By restricting competition from imported goods (illegally) 6.9% Through more liberal imports of inputs 0.4% 4.6% 5.9% 1.0% 0.6% Assistance with raw materials 5.3% 3.6% 2.9% Through assistance with technical know-how 4.7% 3.8% 3.1% By improving training facilities for workers 2.0% 3.0% 4.1% Through assistance with marketing 14.8% 11.1% 6.7% By clarifying sustainable long term government policies 4.1% Through improved macro-economic policies 2.0% 3.3% 3.0% 1.4% 1.8% Through better private sector policies 11.3% 6.9% 5.9% Others 1.6% 0.6% 0.4% Source: Author’s calculation from CIEM, DoE and ILLSA survey from 2005-2013 The results show the optimal state supports for SMEs are providing easier access to credit to SMEs, assistance with land, and removing bureaucratic restrictions In addition, given micro-scale businesses, they need an assistance with marketing and improved macro-economic policies Research methodology 4.1 The method estimated the total factor productivity (TFP) To estimate the TFP, the study begins with the Cobb - Douglas production function of Solow (1957): βk βl (1) Yit = itA it Lit K Yit is the output of a firm i at time t ; Kit , Lit are capital input and labor input respectively and Ait is the effect of firm i at time i Yit , Kit are observed by the econometrics but Ait is an unobserved item Taking natural logs (1) we have a linear production function: yit = β0 + βk kit + βllit + εit (2) ln(Ait ) = β0 + εit measures the average effectiveness of a firm over time ; εit is a deviation from the average value of the characteristics of the manufacturer and the time and εit can be decomposed into an observed (or predictable) component and an unobserved (or unpredictable) component Equation (2) is written into: yit = β k + βl it k it + β l +υ + We have ω = ρ it µ it β +υ , defined as the ρ productivity of firm i at time t and µ error of equation (3) it it (3) is presented for the Next we estimated equation (3) and solve ωit Productivity is estimated as follows: ˆˆ ˆ ˆ ˆ (4) ωit = βo +υit = yit − βkkit − βllit Finally, to estimate the total factor productivity (TFP), we take the log e base of ωˆ it According to Van Beveren (2012), TFP estimation by OLS regression technique is a technically simple However, OLS regression method assumes that the inputs are exogenous but in reality, these elements are endogenous (Olley & Pakes, 1996; Levinsohn & Petrin, 2003) Therefore, the result of TFP estimation is bias To solve the problem for TFP estimation, the estimation techniques such as GMM (Hansen, 1982), OP (Olley & Pakes, 1996) and LP (Levinsohn & Petrin, 2003) are used to estimate TFP in order to avoid bias In this study, we use the estimate TFP by Levinsohn and Petrin (2003) to estimate the total factor productivity at firm level 4.2 Model for testing the learning by exporting mechanism Due to the nature of small and medium enterprises in 2005-2013 the dataset features dynamic panel data and there may be a correlation between independent variables in the model when estimating of the relationship between exporting and TFP That means if we estimate model by OLS regression, it will cause bias to the results To overcome the problem of endogeneity, this study uses GMM regression techniques that were developed by Hasen (1982), Griliches and Hausman (1986), and Arellano and Bond (1991) Model 1: ˆ TFPijt = β + TFP βˆ ˆ + XK2β+ijt ∑ ˆ k ˆ year β + ε ∑ s l k t ˆ l k ˆ β region + + ˆ ˆ ˆ β β β β β sec tor + X∑ X +∑ + X ∑ + X∑ ∑ ijt −1 m n m n l j m j n j j ijt s s t Model 2: TFP + βˆ TFP = β + β ijt + βˆ XK ijt −1 X K ˆ ˆ + XK βˆ ijtChin a ˆ + XK β ijt Russian + βˆ XK ijt Japan ijt Asean + βˆ X + ijt otherAsia k l k + βˆ sec tor ∑+ j jj ˆ β region + ∑ s s s l k β ˆ year ∑ t + ε j ijt t TFPijt : productivity of firm i at industry j at time t TFPijt −1 : productivity of firm i at industry j at time t-1 XKijt : export of firm i at industry j at time t China ijt Asia : export of firm i to China at industry j at time t XKijt Japan : export of firm i to Japan at industry j at time t XKijt Asian : βˆ X + ∑ m m ijt American βˆ X m l Where: XKijt βˆ+ X + βˆ XK + βˆ XK ∑ ∑ ∑ ijt Eupro + βˆ XK n n n export of firm i to Asean at industry j at time t XKijtAsia : export of firm i to Asia at industry j at time t XKijt American : export of firm i to American at industry j at time t XKijt Eupore : export of firm i to Europe at industry j at time t XKijt Russian : export of firm i to Russia at industry j at time t XKijt OtherAsia : export of firm i to other Asia at industry j at time t X k : variables related to characteristic of a firm X l : variables related to owners of a firm X m : variables related to absorptive capacites of a firm X n : variables related to business environments (institution and infrastructure) yearj : year variables sec torj : industry variables regions : region variables 4.3 Description of the variables in the model Table Description of variables Variables Measurement Variables for TFP estimation flation index) log base e of real added Ln (real value added output/in output Log base e real asset value Ln (total asset /inflation Log base index) e employees Ln (employees) Log base e of intermediate Ln (intermediate values/inflati values TFP index on index) (section 2.4.1) Measured by log base Dependent e of ωˆ it variables Group 1: Variables related to firm characteristics and ownerLevel lag of TFP Lag of TFP index index There are three levels of firm size: micro scale Scale of SMEs (employees