Testing Hypothesis H1 and Discussing Research Results

Table 4.13 presents the regression results of the empirical research model with PLS, FEM, REM and REM-RSE. From the regression results of the REM-RSE model, the H-Statistic value of the Vietnamese commercial banking market is determined as H -TTNHVN = 0.543 (0.343 + 0.134 + 0.066). Thus, 0 < H-Statistic < 1, according to the Panzar - Rosse model, this is a case of monopolistic competition. All three input price variables, namely input price of mobilized capital ( w 1 ), input price of employees (w 2 ) and input price of physical capital (w 3 ) have positive and statistically significant regression coefficients. In which, the regression coefficient of the input price of mobilized capital is the highest with 0.343, meaning that a 1% change in the input price of mobilized capital will make the bank's income change by 0.343%. Next is the regression coefficient of the input price of employees with 0.134, meaning that a 1% change in the input price of employees will make the bank's income change by 0.134%, and the lowest is the regression coefficient of the input price of physical capital with 0.066, meaning that a 1% change in the input price of physical capital will make the bank's income change by 0.066%.

The H-Statistic value of the SBV group is H -SBV = 0.169 (0 + (-0.051) + 0.22) reflecting the contribution or influence of the SBV group on H -TTNHVN . In which, the regression coefficient of the input price of mobilized capital is 0, meaning that the change in the input price of mobilized capital has almost no impact on the change in the income of the SBV. The regression coefficient of the input price of employees is less than 0 (-0.051) but is not statistically significant, meaning that the input price of employees also does not affect the income of the SBV. The regression coefficient of the input price of physical capital is statistically significant with a regression coefficient value of 0.22, meaning that a 1% change in the input price of physical capital will cause the income of the SBV to change by 0.22%. The regression results reflect that the income of the SBV group is almost unaffected by the input price of mobilized capital, but only depends on physical capital.

The regression coefficient of the common asset size variable ( ln(AS) ) of the domestic and foreign banks has a value of 1.07 and is statistically significant, showing that if the asset size increases by 1%, the bank's income increases by 1.07% and vice versa, when the asset size decreases by 1%, the bank's income decreases by 1.07%. In which, the regression coefficient of the asset size variable of the foreign banks group ( ln(AS)*D ) is also greater than 0 (0.077) and has

statistical significance

Similarly, the regression coefficient of the loan-to-asset ratio variable ( ln(LO) ) of the domestic and foreign banks is 0.02 and is statistically significant, showing that if lending increases by 1%, the bank's income increases by 0.02% and vice versa, when lending decreases by 1%, the bank's income decreases by 0.02%. However, the regression coefficient of the loan-to-asset ratio variable of the foreign banks group ( ln(LO)*D ) is negative, but not statistically significant. This means that the change in the loan-to-asset ratio does not change the income of the foreign banks. This result is consistent with the analysis results of the input price variable of mobilized capital.

4.2.4. Testing hypothesis H1 and discussing research results

To test hypothesis H1 ( Penetration of foreign banks increases the competitiveness of Vietnam's commercial banking market ), the thesis uses F-test for the value of H -foreign banks estimated from Model 3.2 using the REM-RSE method presented in Table 4.13. The results of testing hypothesis H1 are presented in Table 4.14.

Table 4.14: Results of testing hypothesis H1

H-Statistic	Regression coefficient	Value
H -TTNHVN	( β 1 + β 2 + β 3 )	0.543
H -NHNNg	( β 4 + β 5 + β 6 )	0.169
F-test: H -NHNNg = 0		4.43** (0.035)

Maybe you are interested!

Note: P-values are shown in parentheses. (**) 5% significance level.

Source: Author calculated from research data sample using STATA software.

Through Table 4.14, the results of testing hypothesis H1 show that the H-Statistic value of the Vietnamese commercial banking market is H -TTNHVN = 0.543, in which the contribution of the foreign banks group is H -NHNNg = 0.169, greater than 0 and statistically significant at 95%, which is the basis for accepting hypothesis H1. This shows that the penetration of foreign banks increases the level of competition in the Vietnamese commercial banking market in the period 2009 - 2019.

The research results of this thesis are consistent with some international studies that have been conducted (Cho, 1990; Diallo, 2016; Jeon et al., 2011; Mulyaningsih et al., 2015; Yin, 2020).

Cho (1990) showed that the entry of foreign banks increased the competitiveness of the Indonesian banking market. The results of Jeon et al. (2011) also showed that the presence of foreign banks increased the level of competition in the banking markets of 17 emerging economies in Asia and Latin America.

Recent studies such as Mulyaningsih et al. (2015) also found that foreign bank penetration increased the competitiveness of the Indonesian banking market. Diallo’s (2016) study in 34 African countries, and (Yin, 2020) study in developing countries also showed similar results.

However, the results of this study are completely contrary to the studies of Yeyati and Micco (2007) and Poghosyan and Poghosyan (2010). The results of Yeyati and Micco (2007) showed that the increased presence of foreign banks reduces the level of competition in the banking market in 8 emerging countries in Latin America, or Poghosyan and Poghosyan (2010) argued that the penetration of foreign banks through mergers and acquisitions reduces the level of competition in the banking market of 11 transition economies in Central and Eastern Europe.

The research results on this topic can be explained based on the penetration method of foreign banks. If foreign banks penetrate by establishing new business establishments, it will increase competition in the domestic banking market; if they penetrate by acquiring and merging, it will reduce competition in the domestic banking market. The reason is that when penetrating by establishing new business establishments, foreign banks do not have available customer sources, so they will focus on finding customers (advertising, marketing, service prices, product differentiation, etc.) to gain market share, leading to increased competition; while penetrating by acquiring and merging, foreign banks already have available customer sources, so gaining market share will be less likely. In addition, if the acquisition and merger process of foreign banks

Large-scale mergers will lead to the formation of banking monopolies, and as a result the banking market becomes less competitive.

In Vietnam, foreign banks' penetration is mainly done by establishing new business establishments. During the research period of 2009-2019, 9 100% foreign-owned banks were established and put into operation, the number of foreign banks' branches increased from 40 in 2009 to 49 in 2019. Meanwhile, foreign banks only penetrated by acquiring shares of domestic banks in 5 deals (Common Wealth of Australia bought shares of VIB in 2010, International Finance Corporation bought shares of Vietinbank and Mizuho Bank bought shares of Vietcombank in 2011, Bank of Tokyo Mitsubishi UFJ bought shares of Vietinbank in 2012, KEB Hana bought shares of BIDV in 2019), but there were 7 foreign banks' divestment deals from domestic banks (ANZ bank divested from Sacombank in 2012, OCBC divested from VP bank in 2013, HSBC divested from Techcombank in 2016, Commonwealth Bank of Australia at VIB In 2017, Standard Chartered divested from ACB and BNP Paribas divested from OCB in 2018, Société Générale Group divested from SeaABank in 2019). Thus, the research results of the thesis are consistent with the practical penetration methods of foreign banks in Vietnam.

In summary, the results of this study are the basis for accepting hypothesis H1, answering RQ1 that the penetration of foreign banks increases the level of competition in the Vietnamese commercial banking market. At the same time, the findings of the study demonstrate the consistency with the theory on the impact of foreign bank penetration on increasing the competitiveness of the domestic banking market.

4.3. RESEARCH RESULTS FOR RQ2

This section tests hypothesis H2 as a basis for answering RQ2. As presented in Section 3.2, this study uses a 2-step analysis process: (i) determining the efficiency of domestic commercial banks using 2 methods: financial index method and DEA method, (ii) efficiency indicators will be regressed with variables

foreign bank penetration. The regression coefficients of foreign bank penetration variables are the basis for testing hypothesis H2.

4.3.1. Measuring foreign bank penetration in Vietnam

This section presents the measurement of variables representing foreign bank penetration ( FBA, NFB ) in Models 3.4 and 3.6. The method of measuring foreign bank penetration has been presented in Section 3.2.1, Chapter 3. Accordingly, this study uses a composite method with 2 variables: the ratio of foreign bank assets ( FBA ), and the ratio of the number of foreign banks ( NFB ) (Table 4.15).

Table 4.15: Penetration variables of foreign banks in Vietnam in the period 2009 - 2019

Year	FBA	NFB
2009	3.5%	53.2%
2010	4.5%	58.0%
2011	4.4%	59.6%
2012	4.6%	59.8%
2013	4.2%	62.0%
2014	12.3%	59.6%
2015	10.8%	62.4%
2016	10.2%	62.8%
2017	9.9%	63.2%
2018	10.0%	63.2%
2019	10.0%	63.2%
Medium	7.7%	60.6%
Source: Author compiled from SBV annual report.

The foreign bank asset ratio variable is calculated by dividing the total assets of the foreign bank sector by the total assets of the entire industry. The foreign bank asset ratio has tripled from 3.5% in 2009 to 12.3% in 2014. From 2015 to 2019, the foreign bank asset ratio accounted for 10% to 11% of total assets, down from 1% to 2% compared to 2014. The average foreign bank asset ratio during the study period was 7.7%.

The number of foreign banks is calculated by dividing the number of foreign banks by the total number of banks in the industry. In 2009, the number of foreign banks including foreign bank branches accounted for 53.2% of the total number of banks. By 2016, the number of foreign banks increased by 10% to 63.2%. From 2016 to 2019, the number of foreign banks was stable at 63%. The average number of foreign banks during the research period was 60.6%.

4.3.2. Efficiency of Vietnamese commercial banks by financial index method

To determine the efficiency of Vietnamese commercial banks using the financial index method, this thesis uses the ROA variable as presented in Section 3.2.1, Chapter 3. The average pre-tax profit margin on assets by year in the research period 2009 - 2019 is presented in Table 4.16.

Table 4.16: Average return on assets for the period 2009 - 2019

Year	ROA (%)
2009	1.65%
2010	1.57%
2011	1.30%
2012	1.00%
2013	0.77%
2014	0.68%
2015	0.51%
2016	0.65%
2017	0.71%
2018	0.90%
2019	1.23%
Medium	1.00%
Source: Author calculated from research data sample using STATA software.

The average return on total assets during the study period was nearly 1%. The average return on total assets peaked in 2009 at 1.65%. In the following years, the average return on total assets continuously decreased and reached its lowest level in 2015 at 0.51%. From 2016 to 2019, the average return on total assets tended to increase, reaching 1.23% in 2019.

4.3.3. Efficiency of Vietnamese commercial banks by data envelopment method

To determine the efficiency of Vietnamese commercial banks using the DEA method, the thesis uses the input minimization model and assumes that efficiency varies with scale. The input and output variables in the DEA model are selected according to the intermediate approach with 3 input variables: interest expenses, staff expenses, other expenses, and 2 output variables: interest income and non-interest income according to detailed analysis in Section 3.2.2.2.

The data used in the DEA model includes 30 Vietnamese commercial banks in the period 2009 - 2019 with 315 observations. Descriptive statistics of input and output variables in the DEA model are presented in Table 4.17.

Table 4.17: Descriptive statistics of variables in DEA model

Variable	Value smallest	Value biggest	Central jar	Deviation standard	Number of officials close
Input (billion VND)
Interest expense	139	64,769	9,449	11,839	315
Employee costs	33	14,530	1,693	2,477	315
Other costs	31	11,333	1,645	2,047	315
Output (billion VND)
Earn interest	271	106,468	14,621	19,130	315
Non-interest income	-940	17,274	1,615	2,525	315
Source: Author calculated from research data sample using STATA software.

Table 4.18 presents the average technical efficiency measure of Vietnamese commercial banks during the research period 2009 - 2019.

Table 4.18: Average technical efficiency in the period 2009 - 2019

Year	TE
2009	94%
2010	95%
2011	97%
2012	97%
2013	95%
2014	94%
2015	94%
2016	95%
2017	95%
2018	94%
2019	96%
Medium	95%
Source: Author calculated from research data sample using STATA software.

The average technical efficiency of Vietnamese commercial banks was lowest in the research period at 94% in 2009, then increased from 94% to the highest level of 97% in 2011 - 2012. From 2013 to 2019, technical efficiency continuously fluctuated between 94% and 96%. The average technical efficiency in the research period 2009 - 2019 was 95%. This means that to produce the same level of output, banks used 95% of the inputs, or in other words, banks still wasted about 5% of inputs. The relatively high average technical efficiency shows that Vietnamese commercial banks have focused on effectively managing input resources in the business operation process.

4.3.4. Testing the stationarity of panel data series

This section performs a stationarity check of panel data series to ensure the accuracy of the regression method similar to the analysis steps of Model 3.2 presented in Section 4.2.

Comment