تعیین ساختار سرمایه و بازده سهام - رویکرد LISREL: آزمون تجربی از بازار سهام تایوان

Titman and Wessels (1988) utilize a structural-equations model (LISREL) to find out the latent determinants of capital structure. Maddala and Nimalendran (1996) indicate that the problematic model specification causes the poor results in Titman and Wessels’ research. Chang, Lee, & Lee (2009) apply a Multiple Indicators and Multiple Causes (MIMIC) model to re-examine the same issue as Titman and Wessels did but found more convincing results. We extend Titman and Wessels’ research from using a single-equation approach to a multi-equations approach. In addition to the determinants of firms’ capital structure, those of stock returns are determined simultaneously. Literature indicates that a firm's capital structure may affect its stock returns (Bhandari, 1988), and the reverse is true too (Baker and Wurgler, 2002, Lucas and McDonald, 1990 and Welch, 2004). Hence, a firm's determinants of its capital structure and those of its stock returns should be decided simultaneously, rather than independently. By solving the simultaneous equations, we examine the empirical relationship between the two endogenous variables: capital structure and stock returns and find out their common determinants as well. Our results show that stock returns, expected growth, uniqueness, asset structure, profitability, and industry classification are the main factors of capital structure, while the primary determinants of stock returns are leverage, expected growth, profitability, value and liquidity. The level of debt ratios and stock returns are mutually determined by the aforementioned factors and themselves.

Capital structure and stock returns are topics that have received much attention in the financial management arena. Over the past few years, there are increasing researches surrounding the issues related to the determinants of capital structure and the determining factors of expected stock returns. A few empirical studies combine these two major topics together and test whether debt/equity ratio influences stock returns or whether stock returns are factors of capital structure choices. Bhandari (1988) found that the expected common stock returns are positively related to the ratio of debt to equity and suggested that debt/equity ratio is one of the risk premia of stock returns. Hovakimian, Opler, & Titman (2001) used multiple regressions to explain companies’ leverage degree and concluded that the target debt ratio may change over time as the firm's stock price and profitability change. Lucas and McDonald (1990), Baker and Wurgler (2002) and Welch (2004) empirically show that current capital structure is strongly related to historical market values and that equity issues on average are preceded by abnormal positive stock returns. Although previous researches examined the relationship between capital structure and stock returns, few of them discussed the interaction of the two variables simultaneously. In other words, most of the researches considered only one-way causality: either capital structure affects stock returns or stock returns affect capital structure. In this paper, we try to treat the two variables (debt ratio and stock returns) both as endogenous variables, and use simultaneous linear equations to find out their inter-relationship as well as their own exogenous determinants. Our paper is motivated by Titman and Wessels (1988), who used a structural-equations model (SEM) to extract eight latent variables to explain only an endogenous variable, debt ratio level. We now extend their model to account for two endogenous variable, debt ratio and stock returns, and apply a LISREL system to set up and estimate our models. Our purpose is first to find out the latent factors of firms’ capital structure and stock returns respectively. Secondly, we want to know how firms’ capital structure and stock returns affect each other. Does capital structure affect stock returns or the reverse is true? Or they are mutually affected? If so, which one has more dominant power and how does it work? In this paper, we use Taiwan stock market data on TEJ to examine these hypotheses. We apply a structural-equations model (a LISREL system) to establish our models. LISREL has been widely applied in researches on marketing, organizational behavior, management, sociology, psychology, and accounting as well. Titman and Wessels (1988) first applied LISREL in finance area to deal with the determinants of capital structure. However, few financial researches have used this methodology since then. Maddala and Nimalendran (1996) indicate that the problematic model specification causes the poor results in Titman and Wessels’ research. Chang et al. (2009) apply a Multiple Indicators and Multiple Causes (MIMIC) model to re-examine the same issue as Titman and Wessels did but found more convincing results. In this paper, we try to use a LISREL model to explore more information on the relationship between and the determinants of capital structure and those of stock returns. LISREL, a factor-analytic model, consisting of two parts: a measurement model and a structural model that can be estimated simultaneously. In the measurement model, unobservable firm-specific attributes are measured by relating them to observable variables, e.g., accounting data. In the structural model, measured debt ratio and stock returns are specified as functions of the attributes defined in the measurement models.

In this paper, we introduced a factor-analytic methodology (LISREL) to find out the common determinants of capital structure and stock returns, and also to estimate the impact of these unobservable attributes on the mutual choices of corporate debt ratios and stock returns. A sample of 662-702 Taiwan non-financial companies through year 2003–2005 have been analyzed, using LT/BVA and stock returns or LT/MVA and stock returns as two endogenous variables, and other 11 latent factors were applied to serve as exogenous variables. First, we find that debt levels have positive influences on stock returns whereas stock returns have negative impacts on leverage. These findings are consistent with most empirical studies and may result in a balance condition driving the level of debt ratio and stock return to remain stable within a particular range. Next, we find that two exogenous factors—“profitability” and “expected growth” are the common determinants of debt ratio and stock returns, where they both have negative impacts on leverage level and positive impacts on stock returns. Besides, uniqueness, industry classification and asset structure are three significant determinants only for capital structure, where the first two of them have negative influences and the last one has positive impact on capital structure. For stock returns, in addition to profitability and expected growth, value and liquidity are two other significant factors in determining stock returns, and one has positive influence while the other has negative influence. Our results do not provide support for an effect on debt ratios arising from volatility and an effect on stock returns from size and momentum. Since the structure and methodology of this paper are mainly based on Titman and Wessels (1988) and are extended by adding one more endogenous variable—stock return, it should be helpful to contrast our results. Titman and Wessels investigated the determinants of capital structure using the data of U.S. firms over the 1974 through 1982 time period, while we analyze the common factors of capital structure and stock returns using the data of Taiwan companies through 1998–2005. When we analyze the determinants of capital structure, the signs of the estimates of the two papers are mainly the same, but some of them are not statistically significant. However, we both find significant negative correlations between profitability and the debt ratio measured in market value. The same findings support that the pecking order theory may mostly explain the financing behavior of firms in each market during the according sample years.