“Impact of macroeconomic factors and interaction with institutional performance on Vietnamese bank share prices”

Shares of listed banks in Vietnam gain a lot of interest from investors and regulators. It is important to study the primary drivers of the banks’ share prices. In this context, Gross Domestic Product (GDP), Gold Price (GP), Ninety-day Interbank Interest Rate (R), and USD/VND Exchange Rate (FX) are selected as representatives for macroeconomic variables. A new contribution of this study is the application of interactive fac- tors between macroeconomics and bank performance (i.e., Equity Capital (E), Deposit Аmounts (D), Loan Amounts (L), Non-performing Loans (NPLs), Leverage (LEV), Capital Adequacy Ratio (CAR), Return on Assets (ROA), and Stock Beta (Beta)) in evaluating their impact on bank share prices. Applying the econometric method of Two-Stage Least Square (2SLS) and the quarterly financial data of 13 listed banks from Q1/2009 to Q3/2020, the regression results show that GDP improvements can foster an increase in bank share prices, and this impact is strengthened if banks have good performance of ROA, CAR, and with strict control of NPLs. The R also has a positive impact on bank share prices, and the price level increases if NPLs, LEV, and Beta are controlled at optimal levels. However, empirical evidence drawn from the study also suggests that an increase in FX and GP is not a significant contributor to bank share prices, especially if the bank does not manage NPLs and LEV. Moreover, the impact of E, D, and L on the movements of bank share prices is not significant.


INTRODUCTION
In the security market, news about share prices is published daily, and any change in market price levels can measure stock returns. From the perspectives of both investors and issuers, understanding a series of factors influencing share prices is vital to ensure that investment decisions or equity financing are made at the appropriate time and scale. For Vietnamese commercial banks, raising equity from the private sector, especially from the stock market, is an integral part of their financial strategy. The share prices are affected by a confluence of factors existing in the national economic environment and the creditworthiness of these banks. Determinants of security price behavior can be found in the Dividend Discount Model (Miller & Modigliani, 1961), the Efficient Market Hypothesis (Fama, 1970), and the Arbitrage Pricing Theory (Ross, 1976). Along with the above traditional theories, the dynamic relationship among macroeconomic indicators and share prices are also illustrated in a number of research studies by Fama and French (1989), Jensen and Johnson (1995), King and Watson (1996) . The purpose of this study is to address two important issues: • First, evaluating the level of interactive impact of macroeconomic climate and institutional characteristics of banks on the behavior of bank share prices.
• Second, providing insights for bank regulators, policy makers, and investors in order to assist their understanding of the factors influencing share prices. 1 The VN30 and HNX30 track the performance of the 30 largest stocks on HOSE and HNX by market capitalization and trading liquidity.

LITERATURE REVIEW
The connection between macroeconomic factors and financial asset prices are explained by financial economic theories. According to Keynes (1937), changes in money supply caused changes in the quantity of loanable funds through the banking system, resulting in changes in market interest rates. The Dividend Discount Model developed by Miller and Modigliani (1961) states that the current share price is the sum of expected future cash flow payments; and these are discounted back to present values at an appropriate rate of return and are affected by national economic conditions. Fama (1970) postulated that security prices in efficient capital markets "fully" reflected available information. There are three primary forms of price adjustments: 1) the weak form is associated with historical data of stock prices; 2) the semi-strong form is associated with cur-rent and available information in the public sphere; and 3) the strong form is associated with undisclosed information (non-public) available to powerful investors or business groups.
The Arbitrage Pricing Theory (APT) proposed by Ross (1976) states the prediction of stock returns is based on the estimation of the sensitivity of the stock price combined with systematic factors such as interest rates and foreign exchange rates. Sharpe (1964) developed the Capital Asset Pricing Model (CAPM) and pointed out that the risk-free interest rate, risk bearing interest rates and beta coefficients of the portfolio were the three main drivers for stock prices. Along with the basic CAPM approach, the international CAPM developed by Stulz (1999) highlighted the inclusion of exchange rate in the stock price model. Dornbusch and Fischer (1986) concluded that an appreciation of the domestic currency would produce a positive effect on stock prices of the import-dependent country by reducing costs of production in most business industries and create a negative effect on the domestic stock market for an export-dependent country.
In association with financial economic theories, empirical evidence about the impact of the macroeconomic environment on the behavior of share prices was found ( The link between exchange rates and share prices or between gold prices and share prices is also widely investigated. Garg

METHODOLOGY
The data applied in this study is secondary data provided by FiinPro and General Statistics Office of Vietnam, and the selection of variables is based on the quarter period covering a period from Q1/2009 to Q3/2020. Variables investigated include macroeconomic indicators, the average bank share price at the close of each trading day on HOSE and HNX, and indicators abstracted from the financial statements of 13 listed banks. The exact variables used in the study are listed in Table 1. Except for LEV, Beta and ROA, natural logarithm transformation for other variables is applied with the aim to make the data more symmetric. Two-Stage Least Squares (2SLS) approach is applied to evaluate the impact of macroeconomics and interactive factors between macroeconomics and institutional performance on the bank share prices.
First, the general model using quarterly panel data is suggested as follows: where LnSP it is the natural logarithm of adjusted closing share price of bank i at quarter t; X it is the matrix of macroeconomic variables proposed in Table 1, which represent independent variables; L it is the matrix of interactive variables; and u it is error term.   To investigate the effects of macroeconomics factors and bank performance on bank share prices, four separate econometric models using different combination of interactive variables are specified, in which each of the four key macro-economic variables (i.e., GDP, R, FX and GP) is interacted with the other variables of institutional performance of banks (see Table 2).

RESULTS
The descriptive statistics found in Table 2 indicate that no selected variable is in violation with the condition of collinearity (i.e., VIF value must be lower than 10). Specifically, three variables, including GDP·E, GDP·D, and GDP·L, are removed in Model 1 because of their violation of VIF condition. Similarly, five variables, including R, R·E, R·D, R·L, R·ROA, are removed in Model 2. Three variables, including FX·E, FX·L, and FX·CAR, are eliminated in Model 3, while four variables, including GP·E, GP·D, GP·L, and GP·CAR, are eliminated in Model 4.
The total observations for LnSP are 423 because each bank had different listing times, while those for determinants of bank share prices are 611, ensuring no value is missed during 47 quarters from Q1/2009 to Q3/2020. The average bank share price is 13,586 VND (equal to LnSP = 9.515) associated with a range from 3,465 VND to 90,415 VND (LnSP = 8.149 to 11.410). For the four key macroeconomic variables, the average GDP is reported at 6,539 thousand billion VND and R is equal to 1.962%. Besides, FX in the study period is 9.97 while GP is 17.042 in terms of natural logarithm. Furthermore, details about standard deviation, ranges of determinant variables are shown in Table 2. Meanwhile, R·CAR has a significant and positive relationship with LnSP. In Model 3, GDP and R are significantly positive with LnSP, and regression coefficients are equal to 1.941 and 0.528. Besides, FX·NPL also has a positive and significant relationship with LnSP, and it is equivalent to 0.037. However, results show that GP, FX·L EV, Table 3. Correlations among variables used in the models

Models (1) (2) (3) (4) Variables
LnSP GDP R FX GP -   Wu-Hausman tests are conducted to examine whether or not an endogenous problem exists in the four models. If the P-values in the four models were lower than 5%, then the null hypothesis (Ho) is rejected; indicating the existence of endogeneity. As for results of Sargan tests, P-values in the four models are higher than 5%, implying that the instrumental variables used in the regressions satisfied the required conditions of instrumental variables applied in 2SLS.
Considering the impact of GDP on LnSP, it was found that the GDP variable is positively interacted with bank share prices, and the regression coefficients range from 1.153 to 1.941, indicating that LnSP would increase from 1,153 to 1,941 VND if GDP increases one thousand billion VND. When GDP is interacted with CAR, the impact is stronger and the regression coefficient is equal to 2.014 (see Model 1).
The impact of R on LnSP suggests a positive impact on bank share prices, and regression coefficients are in a range from 0.528 to 1.246, implying that LnSP tends to increase from 528 VND to 1,246 VND if R increases by 1%. The impact is more pronounced if interest rate is interacted with CAR, leading to higher impact level at 2,564 VND.
The impact of FX on LnSP suggests that USD/ VND exchange rates negatively affect bank share prices, and coefficients range from -6.807 to -2.038, meaning that a 1,000 VND increase of USD price in relation to VND might lead to a decrease in bank share prices from -2.038 VND to -6.807 VND. When FX is interacted with D, the impact is also negative, but the level of impact is smaller and regression coefficients range from -1.301 to -0.007.
Finally, the impact of retail gold price on bank share prices also suggests a negative trend. If GP levels are reduced by 1 million VND, LnSP tends to increase by 270 VND to 1,452 VND. If GP interacts with either LEV or ROA, the regression results are negative.

CONCLUSION
During the period from Q1/2006 to Q3/2020, the share price of 13 listed banks in Vietnam responded sensitively to macroeconomic conditions. Sustainable improvement of GDP can support the rise in bank share prices. This impact is weakened or strengthened, depending on how GDP was interacted with characteristics and performance of these banks. As far as positive impact of GDP is concerned, the rise in GDP in association with the rise in ROA, or the rise in GDP in association with CAR would strengthen the positive effect, whereas the rise in GDP in association with NPLs would weaken the positive effect. Specifically, the regression coefficients ranged from 1.153 to1.941 would decrease to 0.016.
Ninety-day interbank interest rate was found to be a key driver for bank share prices. Findings from the study confirmed that a rise in 90-day interbank interest rate leads to a rise in the bank share prices, suggesting higher expectations of investors about returns from bank shares. In association with a good plan to increase the capital adequacy, the impact would be stronger, otherwise, weaker if an increase in interest rate is combined with an increase in non-performing loans, escalating financial leverage and increasing systematic risk of bank shares.
The negative impact of USD/VND exchange rate and the bank share prices in the long run shown in the study suggest that the depreciation of the U.S. currency in the Vietnamese market would result in the fall of bank share price.
The depreciation in retail price of gold would result in a decrease in bank share prices, although statistical significance was not found by its interaction with bank characteristics and performance. This finding is consistent with findings found in the studies of Singhal et al. (2019), Garefalakis et al. (2011) implying that gold is an alternative investment for stocks. Except for capital adequacy, beta, leverage and non-performing loans, evidence about the impact of equity capital, deposit and lending was not found to be statistically significant in this study.
Second, the main goal of this study for determinants of bank share prices in Vietnam did provide a strong empirical basis for decision making by policy makers, regulators and investors. For listed banks, a successful plan to raise equity from the stock market is relevant to the sustainable growth of equity value after fund raising, coinciding with a strong growth of bank share prices in the future. In this regard, taking market interest rates, USD/VND exchange rate and the gold market into consideration and controlling for non-performing loans and the capital adequacy, as well as an appropriate level of financial leverage, are required steps in preparation for equity financing plans. For investors, improving knowledge and skills on analysis of macroeconomic conditions and the bank financial statements, as well as market information, is very important for making accurate decisions. Finally, the State Bank of