THE IMPACT OF GOVERNMENT SUPPORT ON FIRM PERFORMANCE IN VIETNAM: NEW EVIDENCE FROM A DYNAMIC APPROACH

Using a sample of private manufacturing smalland medium-sized enterprises (SMEs) in the period 2007–2015, this paper examines the effect of government support on firms’ financial performance in Vietnam. Contrary to many previous studies, the study finds that government assistance affects firms’ financial performance after controlling for heterogeneity, unobservable factors, and dynamic endogeneity. The finding supports the viewpoint of institutional theory. The study also reveals that assistance measures, such as tax exemptions, soft loans, and investment incentives to promote financial performance, are vital for the development of Vietnamese private SMEs.


INTRODUCTION
Theoretically, the linkage between government support and firms' financial performance cannot be predicted directly by any single theory. Institutional theory emphasises the effectiveness of government subsidies as a catalyst for external investments, and Takalo and Tanayama (2010) show that firms receiving government support may give a positive signal to market-based financiers. As a result, they may receive higher external investment than their counterparts without such support. Also, government support can result in additional funding sources to provide firms with more resources where sources are limited. Furthermore, private enterprises may overcome institutional and other barriers on an uneven playing field through the efficiency of government support (Hansen, Rand, & Tarp, 2009). Consequently, firms with government support will increase research and development (R&D) input and thus improve their performance (Wu, 2017).
On the other hand, rent-seeking viewpoints indicate that government subsidies will not necessarily be distributed effectively because the granting of subsidies is not based on a firm's promising prospects or social contribution. As a result, subsidies based on social networks or political connections are not beneficial to company performance. Such biases in government support tend to increase distortion in the efficient allocation of resources among companies, and hence may result in slow profit growth or the reduction of returns on asset and financial performance (Zhang, Li, Zhou, & Zhou, 2014).
In light of these theoretical perspectives, many empirical studies have been conducted in various countries. However, few studies have focused on the role of government support on the development of small-and medium-sized enterprises (SMEs) in developing countries. In addition, the findings are inconclusive, making it hard to draw general inferences. For example, Fajnzylber, Maloney, and Montes-Rojas (2009) found that government support did not significantly affect profitability in Mexico. However, Hansen et al. (2009) show that government assistance helps firms improve their performance and survival.
The current study differs significantly from previous ones in three ways. First, whereas most studies focus on analysis of the US and other developed countries, this study provides the first evidence of the role of government support on firms' financial performance in Vietnam. Furthermore, different types of government support can have varying effects on firms' financial performance. In our study, we go beyond the extant literature by examining the effect of various types of government support on firms' financial performance. Finally, in methodology, the majority of previous studies (e.g., Zhang et al., 2014) often consider the linkage between government support and firms' financial performance using ordinary least squares (OLS) for pooled or panel data regression. However, such approaches cannot overcome several empirical challenges, such as the endogeneity of explanatory variables. More importantly, the presence of potential dynamic endogeneity can be understood as a firm's past financial performance affecting current performance. Following Wintoki, Linck, and Netter (2012), we overcome these problems by using a two-step system dynamic panel, the generalised method of moments (GMM) model. This paper is structured as follows. The next section provides the background and literature for the research. The following section discusses data sources and analysis framework. Empirical results are presented in the following section. The final section offers a summary and conclusions.

THE BACKGROUND OF GOVERNMENT SUPPORT AND ITS ROLE IN SME PERFORMANCE
Recognising that SMEs, especially private firms, are the critical engine for Vietnamese economic growth, the government of Vietnam has set up supporting measures and issued various decrees. Table 1 lists a series of policy measures, including financial access, human resource development, technical support, and trade and export promotion for SMEs in Vietnam.
Although these policies cover all the various aspects of support for SMEs, difficulties in their implementation still exist because of unclear and unrealistic requirements (Le, 2010). For example, a recent decree (56/2009/ND-CP) lists types of support that SMEs can receive from the government. In practice, however, the guidelines are not clear or lack sufficient detail (Anh, Mai, Nhat, & Chuc, 2011). Consequently, it takes much time and effort for SMEs to receive the support offered. In addition, although the leading role of the state sector has been removed, discrimination against non-state SMEs still exists. In addition, corruption remains widespread (Nguyen & Van Dijk, 2012;Vu, Tran, Nguyen, & Lim, 2018). According to the Central Institute for Economic Management (CIEM, 2010), Vietnamese SMEs are likely to make informal payments for receiving support from the government. Hence, when assessing financial performance, it is not clear whether the benefits of government support outweigh the costs or vice versa. The context motivates us to evaluate whether government assistance is beneficial to the financial performance of firms and if so, how?
The literature has documented many studies considering the linkage between government support and firm performance (Cowling, 2010;Lerner, 1999;Rotger, Gørtz, & Storey, 2012). However, the linkage between government assistance and the performance of SMEs has attracted little empirical attention. On the one hand, some studies show that government support has little effect on SME performance.
For example, using a panel dataset on SMEs in the Japanese manufacturing industry, Honjo and Harada (2006) reveal that government initiatives played an inconsequential role in SME sales, employment, and revenues.
On the other hand, Doh and Kim (2014) explore the effects of governmental policies on SME innovation in regional strategic industries in South Korea using technological development assistance funds as a proxy. Results from empirical models indicate that a positive relationship exists between technological support and innovation performance. The study suggests that governmental financial aid is important for SME innovation. The objective of another study was to analyse the impact of public support on Spanish SME performance, considering technological and economic results. Empirical evidence corroborates the direct, positive influence of support on the technological assets of participants. From the economic performance point of view, economic indicators are positively influenced by the improvement in technological background (Barajas, Huergo, & Moreno, 2017). In some cases, mixed results are found in each study. For example, Morris and Stevens (2010) evaluated the impact of a New Zealand government support programme on participating firms, using a new firm-level panel dataset for 2000 to 2006. They found that the programme achieved significant positive results for sales, although the effect on added value and productivity was less conclusive. Maggioni, Sorrentino, and Williams (1999) examined how the most important government programme to encourage entrepreneurship in Italy affected several aspects of the early performance of new firms. Results showed that the public programme produced mixed effects. Government aid allowed firms to acquire a higher level of technology, but government funding gave rise to entrepreneurial start-ups, which are not always fully efficient.
Few contributions deal with the influence of government support on SME performance in developing countries and these still reach different conclusions. Fajnzylber et al. (2009) consider the role of diverse types of government support on firm performance in Mexico. Their research found that the significant intracountry differences in firm productivity observed in developing economies were due in part to market and government failures that limit the ability of micro-firms to reach their optimal sizes. However, in another article, Wei and Liu (2015) examine the effect of government support in the Chinese context and consider a different type of effect on the innovation performance of firms. They divided government support into what they term "vertical and horizontal support," and adopted an empirical research approach in their study. In their results, the authors highlighted that vertical support, in the form of direct R&D subsidies, horizontal support, and regional innovation policy, have a positive effect on the innovation performance of firms.
In Vietnam, a growing literature examines the role of government support in firm performance. Several studies show that government support is an effective tool to improve firm growth and survival (e.g., Hansen et al., 2009). Other research reveals that the effect of government support on firm performance is negligible or insignificant (Vu, Holmes, Tran, & Lim, 2016). However, the evidence about the linkage between government support and firms' financial performance is little known, especially for private SMEs. In addition, there is limited understanding of the effect of government support on firms' financial performance. Investigating subsidies as a whole instead of types of subsidy may obscure the real effect of government support on firms' performance. More precisely, few studies have examined the relationship between government support and SMEs' financial performance with reference to developing countries, particularly Vietnam, considering the effect of government assistance and types of support on SMEs' financial performance. Hence, the contribution of this study will be to fill the gap in the literature by using a dynamic GMM approach to consider the role of government support on firms' financial performance in the Vietnamese domestic SME manufacturing context.

Data
This study utilises the Small and Medium Enterprise (SME) Survey -Enterprise Development in Vietnam (Copenhagen Centre of Development Research -University of Copenhagen). The surveys were conducted in collaboration with two central Vietnamese partners, i.e., CIEM and the Institute of Labour, Science and Social Affairs (ILSSA). 1 The surveys focused on manufacturing SMEs in Vietnam and were conducted every two years, in 2005, 2007, 2009, 2011, 2013, and 2015 To provide a comprehensive analysis of different types of SMEs, the surveys followed a stratified random sampling method according to ownership structures. The surveys provide a wide range of indicators of firm characteristics, including ownership, industry, enterprise history, government support, types of government support, financial performance, and other information. This dataset made it possible to analyse the impact of government support on the financial performance of Vietnamese SMEs. A common problem with time-variant data is that they are often expressed in current prices. Therefore, our data on current variables are deflated to 1994 prices using GDP deflators to avoid biases that might arise because of inflation. A statistical description of the main variables in our regression estimations is given in Table 2. 2

Econometric models
To quantify the role of government support in firms' financial performance, we apply a dynamic model approach. Such approaches have become increasingly important in recent years to deal with the dynamic nature of economic processes (Flannery & Hankins, 2013). It is this dynamic nature which renders problematic traditional estimation techniques, including OLS and fixed-effects (FE) (Flannery & Hankins, 2013;Wintoki et al., 2012). As shown in many previous studies (e.g., Wintoki et al., 2012), empirical models using firms' financial performance as a dependent variable must be examined in a dynamic framework in which lagged dependent variable(s) are considered as explanatory variable(s).
Technically, the inclusion of lagged dependent variable(s) as independent variables of the empirical models allows researchers to control for unobserved historical factors which have potential influence on current firm performance, in this way reducing omitted variable bias (Wooldridge, 2009). Thus, guided by previous studies (Wintoki et al., 2012), the empirical approach taken in this study is specified below: year dummies industry dummies where Y it is the financial performance (as measured by ROA or ROE) of firm i in year t; α 1 is the estimated coefficient on a one-year lagged dependent variable; government support is widely defined as a dummy variable to reduce measurement errors. This is the main interest variable in the model. In this study, we measure Z is a vector of firm-level explanatory variables used in the model as guided by previous studies (e.g., firm size, firm age, innovation, and leverage). We also control for potential influences arising from differences across industries, using dummy variables for industry classification. μ i represents time-invariant unobserved firm characteristics; ω t denotes time-specific effects which are time-variant and common to all firms. These time-specific effects are captured by year dummy variables; ε it is the classical error term.
The information from the past can be captured sufficiently by two lags of the dependent variable (e.g., Adams & Ferreira, 2009). However, when we ran a specification in which current financial performance is a dependent variable regressed on two lags of past performance and using other covariates as in Equation 1, an insignificant effect of Y it-2 on current firm financial performance was found. This result implies that a one-year lagged dependent variable as an explanatory variable in a first-order autoregressive [AR (1)] structure is enough to control for potential dynamic endogeneity. The specification with AR(1) structure is consistent with the arguments of previous studies (Zhou, Faff, & Alpert, 2014), which show that an AR(1) structure appears to be unavoidable when almost all panel datasets used in corporate finance research are short. Hence, the panel specification model (1) with an AR(1) structure can be written as follow: year dummies industry dummies For the estimation approach, the pooled OLS and the OLS with FE methods will provide inconsistent estimations in the presence of the AR(1) structure and endogenous explanatory variables (Flannery & Hankins, 2013). Some studies use a traditional instrumental variable (IV) approach. However, findings from a set of external instrumental variables seem infeasible when almost no independent variables are considered to be exogenous. Consequently, we use the system generalised method of moments (System GMM) estimator proposed by Blundell and Bond (1998) to correct for this inconsistency and these challenges. This estimator is superior to OLS or fixed effects in controlling for time-invariant unobserved heterogeneity across firms, simultaneity, and dynamic endogeneity (Blundell & Bond, 1998;Wintoki et al., 2012).

EMPIRICAL RESULTS AND DISCUSSION
This section describes the results of the empirical analysis. Table 3 column 1 shows the effect of government support on firms' financial performance when using the OLS approach for pooled data, while Table 3 column 2 shows estimated results after controlling for unobservable time-invariant factors. Table 3 column 3 provides dynamic two-step GMM regressions with basic estimation, while columns 4 to 6 report the results from the estimation of extended specifications.   Table 3 presents the results of the effect of government support on firms' financial performance. Regarding the role of the government support covariate in determining firms' financial performance, pooled data estimations reveal that government assistance has a statistically insignificant influence on ROA. However, the results may be biased because of the absence of control for unobservable characteristics in the model. Attempting to control for time-invariant unobserved features and overcome the challenges noted above, we applied twostep dynamic GMM systems as guided by Wintoki et al. (2012). It should be noted that OLS and fixed effects methods may provide more efficient estimations than the GMM system if explanatory variables are not endogenous. Hence, a Durbin-Wu-Hausman test was implemented for all independent variables as a group if they are actually endogenous. According to Schultz et al. (2010), one-year lagged differences in explained covariates, such as ∆government support it−1 , ∆firm size in log it−1 , ∆Innovation it−1 , Bribe it−1 , ∆Export it−1 , and ∆leverage it−1 , are considered instrumental variables, with year dummies and firm age in log considered as exogenous variables. The results of the test show that the null hypothesis is rejected at the traditional level of significance (1%). The endogeneity of regressors is of concern, so it is necessary to apply the GMM system in this study (Durbin-Wu-Hausman tests for endogeneity of covariates are used).
The results of the specification test are reported in Table 3. A serial correlation test of the AR(2) yields p-values of 0.792 and 0.753. In addition, we determined the validity of the system GMM estimation by applying a Hansen-J test for overidentification. The result is displayed in the last row of Table 3. The p-values of the Hansen-J test are 0.993 and 0.961. These results suggest that the GMM system instrumental variables used in this study are valid. In addition, Table 3 reports the results from an exogeneity test of a subset of our instruments that show a p-value of 0.53 and 0.612. These results suggest that we cannot reject the hypothesis of the exogeneity of the additional subset of instruments used in the GMM system estimates.
Interestingly, a totally different picture emerges when using two-step GMM regression. As reported in Table 3 column 3, the effect of government support on firms' financial performance becomes significant after controlling for unobservable characteristics and dynamic endogeneity. This finding reflects the fact that the results from OLS regression are biased. Specifically, the estimated coefficient of government support shows that firms with government support achieve nearly 4% better financial performance than firms without such support. The positive, significant effect of government support on firms' financial performance is further confirmed in extended specifications and the results are displayed in columns 5 and 6 of Table 3.
With regard to the impact of past firms' financial performance, the estimated results in Table 3 show a positive, significant effect on current performance, when unobservable factors are controlled for by using a dynamic two-step general system. This finding agrees with the empirical results of recent studies (e.g., Wintoki et al., 2012). These results show the importance of controlling for unobservable characteristics and also imply that past firms' financial performance is a vital variable in considering the dynamic nature of the factors affecting current performance. Ignoring this variable in the model can result in researchers failing to capture the real effect of government support on firms' financial performance.
To provide additional insight into the linkage between government support and firm financial performance, this study explores several additional scenarios. First, different types of government support may have various effects on firms' financial performance. Accordingly, this study explores the role of types of government support on firm performance. Interestingly, government technical support for trade activities, training of personnel and technology, has no statistically significant influence on firms' financial performance. However, government financial support has a positive influence on SME financial performance and obviously includes assistance such as tax exemptions, tax reductions, or loans from the Vietnam Development Bank (VDB) or Vietnam Bank for Social Policy with preferential interest rate support. Table 4 shows that exporters tend to achieve better financial performance than nonexporters and this finding is consistent with previous studies (e.g., Vu, Holmes, Lim, & Tran, 2014). Furthermore, the results of columns 3 of Table 4 also show the positive relationship between financial leverage and financial performance covered by the dynamic two-step GMM model when the potential sources of endogeneity and unobservable factors are taken into account. This finding supports the argument of González (2013), who suggests that a firm with higher financial debt may force directors into value-maximising decisions to cope with the higher debt pressure. Consequently, such actions improve firms' productivity and financial performance.
Second, many Vietnamese SMEs are not formally registered and government assistance programmes may depend on whether the firm is registered (Loayza, 1997). Accordingly, the linkage between government support and firms' financial performance is examined further in each sub-group, taking into account the formal status of firms. As one would expect, Table 4 shows that government financial assistance is beneficial for registered but not for unregistered firms. The reason may be that the informality may prevent firms from taking full advantage of government support (Loayza, 1997). In addition, the absence of account books and other required documents hinders unregistered firms from accessing and using these forms of support effectively (CIEM, 2010).  As a final step, the robustness of the results is checked by conducting several scenarios. First, as documented by Wong and Hooy (2018), political connections are typical in countries with weak protection of property rights and in developing countries. In addition, some studies show that our results may be biased, ignoring the role of political connections in investigating the relationship between government support and firms' financial performance (e.g., Zhang et al., 2014). 4 Consequently, in further regressions, a political connection index is added. Furthermore, the measure of firms' financial performance (ROA) is replaced by ROE (return on equity). However, the positive effects of government support on firms' financial performance are still recorded and the results are reported in Table 5.

CONCLUSION AND POLICY IMPLICATIONS
Aiming to contribute to the small but growing amount of empirical evidence concerning the linkage between government support and financial performance, this study contributes to the existing literature by providing the first evidence of the influence on SME financial performance exerted not only by government support but also by types of government subsidy. Based on the empirical results, some of the main findings may be summarised as follows.
Regarding traditional firm characteristic factors, the empirical results are generally consistent with other international empirical studies. For example, exporters who sell in both markets achieve a higher financial performance than non-exporters. In addition, leverage has a positive association with firms' financial performance. Furthermore, it is not surprising that firms marked by corrupt behaviour turn in a lower financial performance than their counterparts that are free of it.
With regard to the connection between government support and firms' financial performance, estimates of the OLS indicate that there is no linkage between the two. However, dynamic two-step GMM estimates reveal that government support has a positive influence on firms' financial performance. Also, GMM approaches show that while financial assistance shows a positive association, technical support proves to be a negative link with firms' financial performance. This suggests that the effect of government support on firms' financial performance varies depending on type of subsidy.
Regarding policy implications, changes in government financial support for firms are accompanied by an improvement in firms' financial performance. This finding implies that private Vietnamese SMEs are often small so that the cancellation of subsidies will have a negative impact on both their growth and financial performance. Our results further show that financial support rather than technical assistance has a positive effect on firms' financial performance. This suggests that it is very important to focus on tax exemptions, interest rate subsidies, and investment incentives since these may help private SMEs improve their growth and financial performance, especially in the present context of discrimination against non-state SMEs.
Vietnam is considered to be a successful example of a transitional economy, having shifted from a centrally planned economy to a market-oriented one with an annual average GDP growth rate of 6.8% during the 1986-2009 period (Le, 2010). Also, according to the World Bank (2012), Vietnam's poverty rate fell from nearly 60% in the early 1990s to 20.7% in 2010. Accordingly, Vietnamese government policy may offer a good example for other transitional economies with similar characteristics and conditions.
There are some limitations to the current study. It uses data from manufacturing SMEs, so its findings may not be representative for other enterprises. In particular, the findings may not be true for large enterprises which command various resources and business approaches, including markets and negotiating power. This suggests that further research on larger firms and other sectors beyond manufacturing should be carried out to draw general conclusions about the relationship between government support and firms' financial performance.

APPENDIx
Definitions and measurements of variables in the models