5. DATA AND METHODOLOGY
5.2. Methodology
According to Fama & French (1992), the meaning of capital structure in financial and non-financial firms is different. While high leverage of non-non-financial firms more likely indicates financial distress, this leverage level is probably normal for financial firms. Flannery (1994) also reports that banks are affected by debt level as the same way as other firms, however, they operate unusually high leverage. Therefore this paper tests the impact of capital structure on performance of non-financial firms separately by eliminating data of financial firms (those with a one-digit standard industrial classification (SIC) code of six). Nevertheless, because of the special role of banks in the traditional bank-centered financial system in Vietnam, this paper will also use independent models to investigate the effect of financial leverage on performance of listed banks in Vietnam.
Moreover, data used in this paper includes the period of one of the worst global financial crises in 2008, which has a large contagion all over the World and Vietnam is no exception.
According to Nguyen et al. (2011), the financial crisis unfolded in 2008 causes the World economy into Great Recession, the contagious effects of the crisis spread over all continents.
Most countries suffered from financial distress and high unemployment rate. Until 2010, Greece and Ireland are considered as the latest victims of the 2008 crisis. In this global context, Vietnamese economy experienced sharp downturn due to the spillover effect of the crisis. Industrial production in the fourth quarter of 2008 declined to 15.6% in comparison with 17.4% in 2007. This period also witnessed a dramatic decrease in Vietnamese GDP. For this reason, this paper also explores the influence of capital structure on performance of Vietnamese listed firms during the downturn period from 2008 to 2010.
5.2.1. Capital structure and performance of non-financial firms
Measure of firm performance
According to Short et al. (2007), ROA, ROE and Tobin’s Q are often used as indicators measuring firm performance. While ROA and ROE are employed to present firm’s accounting performance, Tobin’s Q captures firm’s market performance. ROA is determined as the ratio of earnings after interest and tax divided by total assets, ROE is determined as the ratio of earnings after interest and tax divided by total equity and Tobin’s Q is calculated by dividing total market value of firm (market capitalization + market value of debt) into total asset value of firm (King & Santor, 2008).
Measure of capital structure
Capital structure is captured by using firm’s financial leverage. Based on the study of King
& Santor (2008), leverage ratio (LEV) is calculated by dividing total debt into total equity.
Measure of control variables
According to Saurabh & Anil (2015), high sales growth has a direct effect on profitability of a firm because growth firms can generate more value from investment opportunities. More tangible firms are able to avail more debts due to their collateral value, therefore, tangibility has an impact on firm performance. Size and age are also important factors affecting firm
performance because large size and older firms can take advantage of greater credibility and economies of scale, better capabilities and diversification benefits, which may influence profitability.
Based on the study of Le & Phan (2017), high profitable firms are more efficient and thus expected to have higher performance. Moreover, firms with high level of cash are able to alleviate financial distress problems, more capable of supporting their new projects and paying dividends. Hence, liquidity is assumed to positively correlate with firm performance.
Table 1 provides details of control variables and their measures.
Table 1. Control variables of non-financial firms.
Empirical models
By using the measures of variables above, regression equations for firm performance are formulated as below:
𝑅𝑂𝐴𝑖,𝑡= 𝛽0+ 𝛽1𝐿𝐸𝑉𝑖,𝑡+ 𝛽2𝐺𝑅𝑂𝑖,𝑡+ 𝛽3𝑇𝐴𝑁𝑖,𝑡+ 𝛽4𝑆𝐼𝑍𝑖,𝑡+ 𝛽5𝐴𝐺𝐸𝑖,𝑡+ 𝛽6𝑃𝑅𝑂𝑖,𝑡+ 𝛽7𝐿𝐼𝑄𝑖,𝑡+ 𝜀𝑖,𝑡 (1)
𝑅𝑂𝐸𝑖,𝑡= 𝛽0+ 𝛽1𝐿𝐸𝑉𝑖,𝑡+ 𝛽2𝐺𝑅𝑂𝑖,𝑡+ 𝛽3𝑇𝐴𝑁𝑖,𝑡+ 𝛽4𝑆𝐼𝑍𝑖,𝑡+ 𝛽5𝐴𝐺𝐸𝑖,𝑡+ 𝛽6𝑃𝑅𝑂𝑖,𝑡+ 𝛽7𝐿𝐼𝑄𝑖,𝑡+ 𝜀𝑖,𝑡 (2)
𝑇𝑜𝑏𝑖𝑛′𝑄𝑖,𝑡 = 𝛽0+ 𝛽1𝐿𝐸𝑉𝑖,𝑡+ 𝛽2𝐺𝑅𝑂𝑖,𝑡+ 𝛽3𝑇𝐴𝑁𝑖,𝑡+ 𝛽4𝑆𝐼𝑍𝑖,𝑡+ 𝛽5𝐴𝐺𝐸𝑖,𝑡+ 𝛽6𝑃𝑅𝑂𝑖,𝑡+ 𝛽7𝐿𝐼𝑄𝑖,𝑡+ 𝜀𝑖,𝑡 (3)
Variables Measures
Growth (GRO) The percentage change in sales over year Tangibility (TAN) The ratio between fixed assets and total assets Size (SIZ) Natural logarithm of total assets
Age (AGE) The number of years since incorporation date
Profitability (PRO) The ratio of earnings before interest and taxes to total sales Liquidity (LIQ) The ratio of cash and cash equivalent to total assets
Where 𝛽0 is the intercept of the equations; 𝛽1, 𝛽2, … , 𝛽7 are coefficients of independent variables; 𝑖, 𝑡 specify firm i at time t and 𝜀𝑖,𝑡 is the error term.
5.2.2. Capital structure and performance of banks
To investigate the impact of capital structure on performance of listed banks in Vietnam, the leverage ratio (LEV) is still used to measure capital structure. However, because most of the banks in Vietnam launch the initial public offering (IPO) after 2008, data of market capitalization to calculate Tobin’s Q during the whole period 2008-2016 is lacked. Therefore, only ROA and ROE are used to measure performance of listed banks. Moreover, due to the difference in nature of operation, asset structure and capital structure of banks and non-financial firms; as well as the special role of banks in Vietnamese economy and the mutual impact between banks and macroeconomics, control variables are re-determined.
Based on the study of Siddik et al. (2017), control variables for banks can be categorized into two types: bank-specific variables and macroeconomic variables. This paper also uses growth, size and liquidity as bank-specific control variables, while GDP and inflation factors are considered as macroeconomic variables. The importance of growth, size and liquidity to bank performance is similar to non-financial firms.
Regarding economic factors, Athanasoglou et al. (2008) indicate that during economic downturns, demand for loans may decrease, and thus lower bank profitability. By contrast, during economic booms, when most industries are in good prospects and there are plenty of promising investment projects, demand for loans may increase, which leads to an increase in interest rate and improve bank performance. Therefore, GDP is used as a proxy to measure the impact of economics on bank performance. Inflation also plays a vital role in banking industry. By taking estimated inflation into consideration, banks will adjust their nominal interest rate so that their revenue can override the cost to keep high profits (Siddik et al., 2017).
Measures of bank control variables are described in table 2.
Table 2. Control variables of banks.
The models to capture the impact of capital structure on bank performance are denoted as below:
𝑅𝑂𝐴𝑖,𝑡= 𝛽0+ 𝛽1𝐿𝐸𝑉𝑖,𝑡+ 𝛽2𝐺𝑅𝑂𝑖,𝑡+ 𝛽3𝑆𝐼𝑍𝑖,𝑡+ 𝛽4𝐿𝐼𝑄𝑖,𝑡+ 𝛽5𝐺𝐷𝑃𝑖,𝑡+ 𝛽6𝐼𝑁𝐹𝑖,𝑡+ 𝜀𝑖,𝑡 (4)
𝑅𝑂𝐸𝑖,𝑡= 𝛽0+ 𝛽1𝐿𝐸𝑉𝑖,𝑡+ 𝛽2𝐺𝑅𝑂𝑖,𝑡+ 𝛽3𝑆𝐼𝑍𝑖,𝑡+ 𝛽4𝐿𝐼𝑄𝑖,𝑡+ 𝛽5𝐺𝐷𝑃𝑖,𝑡+ 𝛽6𝐼𝑁𝐹𝑖,𝑡+ 𝜀𝑖,𝑡 (5)
5.2.3. Capital structure and firm performance during financial crisis
To investigate the impact of the financial crisis on the relationship between capital structure and firm performance, a dummy variable - 𝐶𝑅𝐼 is added to the models (1)-(5) above. 𝐶𝑅𝐼 takes the value of 1 for crisis period from 2008 to 2010 and 0 for the post-crisis period from 2011 to 2016. New regression models are denoted as the following:
For non-financial firms
𝑅𝑂𝐴𝑖,𝑡= 𝛽0+ 𝛽1𝐿𝐸𝑉𝑖,𝑡+ 𝛽2𝐺𝑅𝑂𝑖,𝑡+ 𝛽3𝑇𝐴𝑁𝑖,𝑡+ 𝛽4𝑆𝐼𝑍𝑖,𝑡+ 𝛽5𝐴𝐺𝐸𝑖,𝑡+ 𝛽6𝑃𝑅𝑂𝑖,𝑡+ 𝛽7𝐿𝐼𝑄𝑖,𝑡+ 𝛽8𝐶𝑅𝐼 + 𝜀𝑖,𝑡 (6)
Variables Measures
Bank-specific
Growth (GRO) The percentage change in sales over year
Size (SIZ) Natural logarithm of total assets Liquidity (LIQ) The ratio of cash and cash equivalent
to total assets
Macroeconomic Gross Domestic Product (GDP) Natural logarithm of GDP Inflation (INF) The annual inflation rate
𝑅𝑂𝐸𝑖,𝑡= 𝛽0+ 𝛽1𝐿𝐸𝑉𝑖,𝑡+ 𝛽2𝐺𝑅𝑂𝑖,𝑡+ 𝛽3𝑇𝐴𝑁𝑖,𝑡+ 𝛽4𝑆𝐼𝑍𝑖,𝑡+ 𝛽5𝐴𝐺𝐸𝑖,𝑡+ 𝛽6𝑃𝑅𝑂𝑖,𝑡+ 𝛽7𝐿𝐼𝑄𝑖,𝑡+ 𝛽8𝐶𝑅𝐼 + 𝜀𝑖,𝑡 (7)
𝑇𝑜𝑏𝑖𝑛′𝑄𝑖,𝑡= 𝛽0+ 𝛽1𝐿𝐸𝑉𝑖,𝑡+ 𝛽2𝐺𝑅𝑂𝑖,𝑡+ 𝛽3𝑇𝐴𝑁𝑖,𝑡+ 𝛽4𝑆𝐼𝑍𝑖,𝑡+ 𝛽5𝐴𝐺𝐸𝑖,𝑡+ 𝛽6𝑃𝑅𝑂𝑖,𝑡+ 𝛽7𝐿𝐼𝑄𝑖,𝑡+ 𝛽8𝐶𝑅𝐼 + 𝜀𝑖,𝑡 (8)
For banks
𝑅𝑂𝐴𝑖,𝑡= 𝛽0+ 𝛽1𝐿𝐸𝑉𝑖,𝑡+ 𝛽2𝐺𝑅𝑂𝑖,𝑡+ 𝛽3𝑆𝐼𝑍𝑖,𝑡+ 𝛽4𝐿𝐼𝑄𝑖,𝑡+ 𝛽5𝐺𝐷𝑃𝑖,𝑡+ 𝛽6𝐼𝑁𝐹𝑖,𝑡+ 𝛽7𝐶𝑅𝐼 + 𝜀𝑖,𝑡 (9)
𝑅𝑂𝐸𝑖,𝑡= 𝛽0+ 𝛽1𝐿𝐸𝑉𝑖,𝑡+ 𝛽2𝐺𝑅𝑂𝑖,𝑡+ 𝛽3𝑆𝐼𝑍𝑖,𝑡+ 𝛽4𝐿𝐼𝑄𝑖,𝑡+ 𝛽5𝐺𝐷𝑃𝑖,𝑡+ 𝛽6𝐼𝑁𝐹𝑖,𝑡+ 𝛽7𝐶𝑅𝐼 + 𝜀𝑖,𝑡 (10)