Credit Rating Levels of Developed Economies in the 2013

and short-term capital. This ratio reflects the ability to use highly liquid assets such as cash, deposits at other commercial banks, etc. to meet the demand for payment of mobilized deposits and short-term capital sources. The higher this ratio, the less liquidity risk the commercial bank faces.

Banks with high liquidity are less likely to go bankrupt than other commercial banks. In addition, Shen et al. (2012) also demonstrated that the ratio of highly liquid assets/total mobilized capital, deposits and short-term capital has a positive impact on MXHTN.

(+)

Maybe you are interested!

LiAss_Debt

The ratio of highly liquid assets/total mobilized capital, deposits and loans. Similarly, commercial banks with high ratios of this ratio are less likely to face liquidity risks. However, if this ratio is maintained at too high a level, it will affect the profitability and operating efficiency of the bank.

Commercial Bank.

Ioannidis et al. (2010) demonstrated that the ratio of highly liquid assets/total mobilized capital and borrowed capital has a positive impact on MXHTN.

(+)

Note: +: positively correlated with dependent variable.

-: negatively correlated with the dependent variable.

*: Classification of commercial banks according to ownership form by Berger et al. (2007)

apply.

3.2 Research data

The research data of the thesis is in the form of cross-sectional data on the MXHTN of commercial banks, financial indicators of these units and macro factors related to the operating environment that affect the MXHTN of commercial banks.

According to data published on Fitch's website as of June 2015, this organization has conducted MXHTN assessments for 650 commercial banks in emerging economies. The author uses the following formula to determine the sample size for the thesis:

(Watson, 2001)

In there:

n: number of observation samples to select.

N: number of objects in the entire set.

Z: is determined based on the required reliability of the model. The author chooses 95% reliability corresponding to Z =1.96.

R: the rate of information collected from observations. For financial statements of commercial banks rated by Fitch, the expected rate of information collection is 100%.

P: variation of the attribute of the object to be observed. The purpose of the model is to survey the MXHTN of commercial banks, the author chooses the average level corresponding to P

=0.5.

A: the accuracy level of the forecast data from the model (0.03; 0.05; 0.1 correspond to the accuracy level of 3%, 5% or 10%). The author decided to choose the average accuracy level of 5%. According to the above formula, the minimum number of observations needed for the thesis's observation sample is n = 242.

To achieve the research objectives of the thesis, the author decided to choose 2 research data samples for the thesis. Data sample 1 includes observations of MXHTN and factors affecting MXHTN of commercial banks in developed economies. Data sample 2 includes observations of MXHTN and factors affecting MXHTN of commercial banks in emerging economies.

According to the article World Economic Outlook 2014 (IMF, 2014), the number of developed economies is 36 and the number of emerging economies is 153. The number of developed economies in the thesis selected by the author is 22 countries including the G7 group and other economies selected from the remaining 29 economies in the group of developed economies by systematic sampling method with a step of 2. The number of emerging economies in the thesis is 41 countries, these economies are selected by systematic sampling method with a step of 3 from the set of 153 economies.

emerging economies and after the authors excluded countries without data on commercial banks in the Bankscope data source.

Next, the author uses a step-by-step sampling method to select observations for the data sample of commercial banks in developed economies and the data sample of commercial banks in emerging economies from the list of commercial banks in the countries identified in the above analysis step in the Bankscope data source. Specifically, the data of the thesis consists of 2 small data samples: data sample 1 includes 296 observations of commercial banks in developed economies, data sample 2 includes 282 observations of commercial banks in emerging economies. The number of observations in both data samples of the thesis meets the minimum number of observations for the observation sample calculated in the above section. Details of the number of commercial banks by country in the 2 sub-data samples of the thesis are presented in detail in Appendix 1a and Appendix 1b.

Data on the MXHTN of commercial banks are taken from Fitch's MXHTN announcements for the period 2013 - 2015. Data on financial indicators of commercial banks are collected for the period 2010 - 2014 from the Bankscope data source. Fitch, Standard & Poor's and Moody's and a number of other credit rating agencies all use data from Bankscope in assessing the MXHTN of commercial banks (Poon and Firth, 2005). Systemic factors related to the operating environment of commercial banks such as: long-term credit rating of the country where the bank is headquartered and the operational risk assessment of the banking industry in these countries are taken from the announcements of the credit rating agency Fitch.

The reason the author limited the time of credit rating information in the period 2013 - 2015 is because if the research time range is expanded, the MXHTN assessment standards of credit rating organizations may change. The study by Savador et al. (2014) demonstrated that the decline in MXHTN of commercial banks in Spain in the period 2000 - 2009 was not only due to instability in the financial situation of commercial banks but also due to the tightening of assessment standards by credit rating organizations. In fact, in the period from 2003 to 2014, Fitch issued 3 different sets of rules for assessing MXHTN of commercial banks: Bank rating methodology in 2003, Global financial institution rating criteria version 2012 and 2014.

Similar to the studies of Matousek and Stewart (2009), Bellotti et al. (2011a, 2011b), Caporale et al. (2012), the financial indicators of commercial banks in the thesis are taken in year t-1 compared to the time when the MXHTN of commercial banks is announced. Because, credit rating agencies evaluate the MXHTN of commercial banks based on the available information of commercial banks at the time of evaluation.

Table 3.2: Credit rating of commercial banks in developed economies in the research data sample for the period 2013 - 2015

Credit rating

Quantity	Proportion
AAA	11	3.7162%
AA	48	16.2162%
A	152	51.3514%
BBB	52	17.5676%
BB	25	8.4459%
B	8	2.7027%
Total	296	100%

Source: Author's calculations from research data sample

Table 3.3: Credit rating of commercial banks in emerging economies in the research data sample for the period 2013 - 2015

Credit rating

Quantity	Proportion
A	32	11.3475%
BBB	116	41.1348%
BB	64	22.6950%
B	70	24.8227%
Total	282	100%

Source: Author's calculations from research data sample

Table 3.2 and Table 3.3 show the number and percentage distribution of credit ratings of commercial banks in the two data samples of the thesis. Thereby, we can see that commercial banks in countries belonging to the group of emerging economies only receive credit ratings from A and below. Meanwhile, AAA and AA credit ratings are all from commercial banks in countries with developed economies. In addition, the credit rating of commercial banks accounts for the largest proportion in countries with developed economies, which is credit rating A (51.35%), while in countries belonging to emerging economies, the credit rating of commercial banks accounts for the largest proportion, which is credit rating BBB (41.13%). From that, we can see that there is a certain impact of the economic characteristics of a country on the credit rating of commercial banks in that country.

3.3 Research hypotheses

Based on the analytical framework in chapter 2, combined with a review of empirical studies related to the MXHTN of commercial banks as well as referring to the MXHTN assessment methods of international credit rating organizations, the author finds that the factors affecting the MXHTN of commercial banks include two main groups of factors: a group of systemic factors and a group of factors reflecting the characteristics of commercial banks. In the group of systemic factors, the author focuses on analyzing the differences in the impact of the level of national risk and the level of risk of the banking industry on the MXHTN of commercial banks in developed economies compared to emerging economies. In the group of factors reflecting the characteristics of commercial banks, the thesis analyzes the differences in the impact of ownership characteristics, total asset size and financial indicators on the MXHTN of commercial banks in developed economies compared to emerging economies. Therefore, the author successively develops the following hypotheses to answer the research questions and thereby achieve the research objectives of the thesis.

As financial intermediaries and important capital channels in the economy, commercial banks are directly and strongly affected by macroeconomic policies as well as general risks of the economy. In addition, commercial banks are also units that are greatly affected by the industry environment and are easily affected by chain effects (Williams et al., 2013). On the other hand, empirical studies on the MXHTN of commercial banks such as those of Matousek and Stewart (2009), Bellotti et al. (2011a, 2011b), Caporale et al. (2012) all show that the risk level of the country where the commercial bank is headquartered has a strong impact on the MXHTN of commercial banks in developed and emerging economies. Similarly, the study by Borensztein et al. (2013) demonstrates that there is a significant impact of a country's credit rating ceiling on the MXHTN of economic organizations in that country. In contrast to the conclusions about the important influence of country risk factors on the MXHTN of commercial banks in the above studies, Poon et al. (1999) pointed out that the level of country risk and operating environment factors have no impact on the MXHTN of commercial banks. In addition, Liu and Ferri (2001) also argued that country risk has no significant impact on the MXHTN of enterprises in OECD countries. However, in countries

In countries that are not part of the OECD group, the country risk factor has a great influence on the MXHTN of enterprises in these countries. Similarly, Moody's (1999) also asserted that the state of the economy and the operating environment have a stronger impact on the MXHTN of commercial banks in emerging economies than in developed economies. Therefore, the following hypothesis is built to test:

Hypothesis 1 (H1) : There is a difference in the impact of the country risk level and the operational risk assessment level of the banking industry where the commercial bank is headquartered on the social capital of commercial banks in developed economies compared to emerging economies.

According to the description in the credit rating method of Fitch and Standard & Poor's, the MXHTN of state-owned commercial banks or those owned by corporations can be improved thanks to the support of the government or parent corporation. Specifically, the government or parent corporation often tends to support capital for commercial banks owned by these organizations when necessary. In addition, the study of Iannotta et al. (2010) proves that large state-owned commercial banks in Europe often have better bond risk ratings than privately owned commercial banks. In addition, the study of Liu and Ferri (2001) also shows that the size of state-owned enterprises or large foreign corporations has a positive impact on the MXHTN of enterprises. However, there are also studies that show that the state or parent ownership factor has a negative impact on the risk level and performance of commercial banks in emerging economies. Specifically, Lang and So (2002) argue that government-owned commercial banks in emerging economies often have bureaucracy in management. Because the boards of directors in these units are often not controlled as effectively as in private commercial banks. Therefore, these individuals do not really make efforts in management or may act for their own benefit. Therefore, state-owned commercial banks in emerging economies often have higher risk levels and operate less effectively than private commercial banks. In addition, Lassoued et al. (2016) argue that foreign investors participating in ownership of commercial banks in emerging economies can increase the risk level for these commercial banks. Because, foreign investors may be more risk-averse in banking activities than domestic investors. Because foreign investors have

can easily diversify their risk exposure internationally. Therefore, the following hypotheses are formulated to test:

Hypothesis 2 (H2) : There is a difference in the impact of the ownership factor of large-scale and prestigious international financial corporations in commercial banks on the social capital of these units in developed economies compared to emerging economies.

Hypothesis 3 (H3) : There is a difference in the impact of government ownership of commercial banks on their corporate social responsibility in developed economies compared to emerging economies.

According to the theory of returns to scale, commercial banks with large total assets can create certain advantages in terms of capital mobilization and reduction of fixed costs compared to commercial banks with smaller total assets. Empirically, studies by Ioannidis et al. (2010), Matousek and Stewart (2009), Bellotti et al. (2011a, 2011b) all demonstrate that total asset size has a positive impact on the MXHTN of commercial banks in emerging and developed economies. On the contrary, Köhler (2015) argues that commercial banks with large total assets in emerging economies tend to operate less stably than commercial banks with small total assets. Because these commercial banks often have diversified business activities and these units use the advantage of diversification to compensate for maintaining a low equity/total asset ratio. In addition, according to Lassoued et al. (2016), large-scale commercial banks in emerging economies are often controlled by the government in lending to projects serving the country's development goals and policies. At this time, commercial banks often do not pay attention to the economic efficiency and the ability to repay the projects because these units "rely" on the government's support when risks occur. Based on the conflicting conclusions about the impact of the total asset size factor on the risk level and operating efficiency of commercial banks mentioned above, the following hypothesis is built to test:

Hypothesis 4 (H4) : There is a difference in the impact of total asset size on the liquidity of commercial banks in developed economies compared to emerging economies.

Next, the study of Shen et al. (2012) demonstrated that in countries with developed economies, the impact of financial indicators on

The impact of financial indicators on the MXHTN of commercial banks is clearly demonstrated. On the contrary, in countries belonging to the group of emerging economies, the impact of financial indicators on the MXHTN of commercial banks is unclear. In addition, the study of Poon and Firth (2005) also demonstrated that the impact of some financial indicators on the MXHTN of commercial banks is different between commercial banks that proactively request to be assessed for MXHTN in developed economies compared to commercial banks that do not proactively request to be assessed for MXHTN in emerging economies. However, the research results of Roy (2005) demonstrated that the impact of some financial indicators on the MXHTN of commercial banks is the same regardless of whether the commercial bank proactively requests to be assessed for MXHTN or not. Similarly, Purda (2003) also found that there is no significant difference in the impact of financial indicators on the MXHTN of enterprises in different countries. Therefore, the author puts forward the following hypothesis to test:

Hypothesis 5 (H5) : There is a difference in the impact of financial indicators on the MXHTN of commercial banks in developed economies compared to emerging economies.

3.4 Data analysis methods

The sequence of steps in data analysis of the thesis to achieve the research objectives and answer the proposed research questions is summarized in the following diagram:

Assess the relevance and test the assumptions in

model

Identify factors that affect the social network of

Commercial Bank

Merge data samples and add interaction variables

Identify differences in the impact of factors affecting social media

Diagram 3.1: Sequence of the thesis analysis steps

One-factor variance analysis of indicators

finance




Select explanatory variables in the ordered logit model

Source: Author's synthesis from theoretical review and related studies.

Step 1, to determine the difference in the impact of factors affecting the MXHTN of commercial banks in developed economies compared to emerging economies, we must first specifically identify which factors affect the MXHTN of commercial banks.

Commercial banks in developed economies and emerging economies. To achieve this goal, the author combines the One-way ANOVA method and the method of selecting explanatory variables in the Ordered logit model on the data sample of commercial banks in developed economies and emerging economies separately.

One-way ANOVA helps to identify explanatory variables that can differentiate between commercial banks belonging to different social networks 3 . This method was applied by Boyacioglu et al. (2009) in the study to select explanatory variables to include in the model. However, the one-way ANOVA method is performed based on some important assumptions such as: the variance of the comparison groups must be homogeneous, the comparison groups must have a normal distribution or the sample size must be large enough to be considered as asymptotic to the distribution

standards, ... However, financial indicators in real research data often find it difficult to meet these assumptions. Therefore, the author conducts the Kruskal - Wallis non-parametric test to support the 1-factor variance analysis method.

The method of selecting explanatory variables in the Ordered logit model used by the author in this thesis has been used by researchers such as: Malhotra and MalhotraRoy (2003), Jardin (2010), Ioannidis et al. (2010) in their empirical studies. This method aims to select appropriate explanatory variables for the Ordered logit model, while also ensuring that the built model does not overfit the research data sample. Accordingly, the author selected 5 sub-data samples from the original data sample. Each sub-data sample has a number of observations equal to 80% of the number of observations in the original data sample. Sub-data samples are selected from the original data sample using the systematic sampling method with a step of 5. According to this sampling method, the author will determine a random number x in the range from 1 to 5. Next, the author will select elements in order x, x+5, x+10,… in the original sample set to form the sub-data sample. After that, the author performs 5 separate model estimates on each different sub-data sample. To determine the explanatory variables that have the main impact on the dependent variable in each specific Ordered logit model, the author applies the backward elimination method.

3 The contents of the one-way ANOVA method are presented in detail in Appendix 5.

stepwise). In this way, the explanatory variables in the model are gradually eliminated based on the statistical significance of the individual regression coefficients. This elimination process starts with the explanatory variable with the least statistically significant individual regression coefficient. Then, the Ordered logit model is re-estimated with the remaining explanatory variables. The elimination process continues until all the individual regression coefficients of the remaining explanatory variables in the model are statistically significant at the 10%, 5%, or 1% level, or until the elimination of additional explanatory variables from the model makes the regression coefficients of the remaining explanatory variables in the model no longer statistically significant.

The statistical significance of the regression coefficients and the frequency of occurrence of explanatory variables in the regression models built on the above sub-data samples and the entire data sample are the basis for determining the explanatory variables that affect the dependent variable according to the method of selecting explanatory variables for the Ordered logit model. Accordingly, explanatory variables with a frequency of occurrence of 50% or more in the Ordered logit models built on the sub-data samples and the entire data sample are considered the most suitable explanatory variables on the original data sample.

The explanatory variables that have the main impact on the MXHTN of commercial banks in developed economies and in emerging economies must be covariates that are capable of distinguishing between commercial banks belonging to different MXHTN as determined by the one-factor variance analysis method and the most suitable explanatory variables on the original data sample are determined from the explanatory variable selection method in the above Ordered logit model.

In the second step, the author uses the BIC (Bayesian information criteria) to evaluate the suitability of the Ordered logit model built on the explanatory variables selected from the above analysis step compared to the Ordered logit models built on other sets of explanatory variables that can be randomly selected from the observed data sample. In addition, the author also tests the assumptions in the Ordered logit model such as: testing for multicollinearity, testing for heteroskedasticity, and testing for the omission of necessary explanatory variables in the model.

Step 3, to achieve the first and second research objectives, namely to determine the difference in the impact of factors affecting the MXHTN of commercial banks in developed economies compared to emerging economies, the author pooled data samples.

commercial banks in developed economies with a sample of commercial banks in emerging economies. At the same time, the dummy variable Emer is added, which has a value of 1 if the commercial bank is headquartered in an emerging economy, and 0 otherwise. Then, the author constructs interaction variables between the variable Emer and each explanatory variable representing systemic factors such as country risk level, banking industry risk level, and factors representing the specific characteristics of each commercial bank such as size, ownership characteristics, and financial indicators in the model. Finally, the author re-estimates the Ordered logit model with the explanatory variables identified from step 1 on a data sample including commercial banks in emerging economies and commercial banks in developed economies, and adds the above interaction variables. In case the regression coefficient of the interaction variable is statistically significant, this proves that there is a difference in the impact of the corresponding explanatory variable on the MXHTN of commercial banks in developed economies compared to the MXHTN of commercial banks in emerging economies. In this case, the impact of each specific factor on the MXHTN of commercial banks is determined by the combined impact of the explanatory variable representing this factor and the interaction variable between the above explanatory variable and the dummy variable Emer. This method has been used by Berger et al. (2010), Shen et al. (2012), Mirzaei et al. (2013) in their studies. Specifically, if the regression coefficient of the interaction variable between an explanatory variable and the Emer variable is statistically significant and the regression coefficient of the interaction variable has the same sign as the regression coefficient of this explanatory variable, the impact of the above explanatory variable on the MXHTN of commercial banks is enhanced in the case of commercial banks headquartered in emerging economies compared to the case of commercial banks headquartered in developed economies. Conversely, when the regression coefficient of the interaction variable has the opposite sign to the regression coefficient of the corresponding explanatory variable, the impact of this explanatory variable on the MXHTN of commercial banks is reduced in the case of commercial banks headquartered in emerging economies compared to the case of commercial banks headquartered in developed economies. In both cases, the author can conclude that there is a difference in the impact of each specific factor on the MXHTN of commercial banks in developed economies compared to emerging economies. On the contrary, the author can conclude that there is no difference in the impact of each specific factor on the MXHTN of commercial banks in the two groups of countries mentioned above. This analysis process is summarized in the diagram below.

Figure 3.2: Identifying the differences in the impact of factors affecting the MXHTN of commercial banks in emerging economies compared to developed economies

Pooling commercial bank data samples from developed and emerging economies

Add the dummy variable Emer to the pooled data sample

Let the dummy variable Emer interact with each explanatory variable to be studied on the pooled data sample.

Re-estimate the model with the explanatory variables identified from step 1 and add the interaction variables to be studied.

The regression coefficient of the interaction variable is not statistically significant.

The regression coefficient of the interaction variable is statistically significant.

There is no difference in the impact of the explanatory variable under study on the MXHTN of commercial banks in emerging economies compared to developed economies.

There are differences in the impact of the explanatory variables to be studied on the MXHTN of commercial banks in emerging economies compared to developed economies.

interaction variable u has a (+) sign

a interaction variable has a (-) sign

Regression coefficient of the research need

The impact of the explanatory variable under study on the MXHTN of commercial banks is enhanced when commercial banks are headquartered in emerging economies compared to developed economies.

The regression coefficient to be studied

The impact of the explanatory variable to be studied on the MXHTN of commercial banks decreases when commercial banks are headquartered in emerging economies compared to developed economies.

Source: Summary from author's analysis steps

The process of assessing the combined impact of the above-mentioned interaction variables and explanatory variables is applied by the author in turn and separately to each factor affecting the MXHTN of commercial banks such as: systemic factors and factors showing the specific characteristics of commercial banks including scale, ownership characteristics and financial indicators.

Credit Rating Levels of Developed Economies in the 2013 - 2015 Research Data Sample

Comment