在最近的文献中,一些研究已经进行了是否影杠杆作用意味着恢复或者没有恢复的测试。经验主义的主要焦点是在速度的调整以及决定目标速度调整的因素。然而文献普遍认同杠杆展览意味着均值回归,这在文献中没有达成关于速度调整的共识。
进行传统杠杆均值测试的有Taggart(1977)、Marsh(1982)和Auerbach (1985)。这些早期研究需要一个长期的平均数作为目标并假设在样本期内影响杠杆的因素维持不变。Marsh(1982)在他的样本周期中建立了周期平均比率目标并通过分对数模型,发现发行债券或股票在目标中负债资产比率偏差的概率。
更多最近的研究有 Flannery and Rangan (2006) , Hovakiam et al. (2001), Kayhan and Titman (2007),Fama and French (2002)。他们发展了调整模型,就是利用杠杆效率随着目标进行及时的调整。这个模型在展示公司时变和非时变特征时都起到了关键的作用。
资本结构的决定因素和速度的调整-Determinants Of Capital Structure And Speed Of Adjustment
In the recent literature, several studies have been conducted to test if the leverage is mean –reverting or not. The main focus of the empirical studies is on the speed of adjustment and the factors that determine the speed of adjustment to the target. While the literature commonly agrees that leverage exhibits mean reversion, there is no consensus in the literature about the speed of adjustment.
The traditional tests of mean reversion are the ones of Taggart (1977), Marsh (1982) and Auerbach (1985). This early studies take a long term average as a target and assumes that the factors that influence the leverage remain unchanged over the sample period. Marsh (1982) establish the target to be the average ratio for the period taken in his sample period and by means of a logit model and find that the probabilities of issuing debt or equity vary with the deviation of the debt-equity ratio from the target.
The more recent studies are the ones of Flannery and Rangan (2006) , Hovakiam et al. (2001), Kayhan and Titman (2007),Fama and French (2002).They developed adjustment models, in which leverage adjust in time towards the target. The model is a function of both time-variant and time-invariant firm characteristics.
Two approaches have been used in the recent literature to estimate the partial adjustment:
Some studies estimate the speed of adjustment in a single step which is the reduced form of the two step procedure: Flannery and Rangan (2006).
Other studies estimate the model in two steps: Fama and French (2002), Hovakiam et al. (2001),Korajczyk and Levy (2003). Firstly, the target is computed externally based on the historical data or estimated based on the proxies documented in the literature to influence the leverage and then use it as an independent variable in the second step which is the partial adjustment regression.
Fama, Eugene F., and Kenneth R. French (2002), Hovakimian et al. (2001) tested if the leverage is mean-reverting as the trade-off theory predicts and if financing decisions respond to short-term variation in earnings and investment as the pecking order predictions states. Their results show that the leverage is mean reverting but that the speed of adjustment is very low (from 7 to 17 percent per year). The rate of reversion was characterized by Fama and French (2002) as a “snail’s pace”. On the other hand, there are studies that documented a fast speed of adjustment. The results of Flannery and Rangan (2006) were that companies do have a target ratio and that typical firm converges toward its long-run target at a rate of more than 30% per year after controlling for the fixed effects. They state “that targeting behavior explains far more of the observed changes in capital structure than market timing or pecking order consideration” [1] .Alti (2006) also finds a much faster adjustment of leverage.
In order to test if the European firms maintain a target leverage as the trade-off theory predicts I will adopt a model with partial adjustment toward target leverage depending on the firm’s variables that have been documented in the literature to determine the capital structure. This procedure has been used in previous studies like the ones of Hovakiam et al (2001), Fama and French (2002) and Flannery and Rangan(2006).
A fast speed of adjustment is consistent with the trade-off theory: that the firms adjust their leverage ratio to the target leverage while a low or non-existent speed of adjustment is a support for the opposite theories that suggest that the leverage is not mean-reverting. Hence, the speed of adjustment is believed to differentiate among the competing capital structure theories.
The model considered implies an unbalanced panel data set that allows the target leverage to vary both across time and across firms. In the next specification of the model, t represents the specific number of reported years and i is the firm's index. First, the target leverage is expressed as a function of the variables that are expected to influence the leverage:
Where is the target leverage at t+1
are the explanatory variables that influence the target leverage
After establishing the target leverage, a standard partial adjustment model can be presented:
Where is the speed of adjustment to the target leverage
is the difference between the actual leverage at t+1 and the actual leverage at t
is the difference between the target leverage at t+1 and the actual leverage at t
By substituting equation (1) in equation (2), following the study of Flannery and Rangan (2006), the model for estimating the speed of adjustment will result:
In equation (3) the speed of adjustment is given by the difference between one minus the coefficient of the lagged variable that will be estimated through the econometrical method. It suggests that in the case that firms adjust towards the target leverage, they will close the gap between the level of the target leverage that they have (which is denoted by and the level at which they want to be:). So, if the firms have a targeting behavior the value of will be closed to 1 that would result in a complete adjustment toward the optimal level, implying that there that the adjustment is not costly. If >0, there will be a partial adjustment toward the optimal level, the firms do not fully adjust their leverage ratio from the previous period to the current one. If there is no adjustment, and consistent with the pecking order theory, firms follow there financing order preferences regardless of shocks.
Hypothesis: the speed of adjustment >0, as the dynamic trade-off model predicts
In order to test the effects of the macroeconomic, institutional and market variables on the speed of adjustment I will adopt the model of Drobetz and Wanzeried (2004), and endogenize the speed of adjustment by writing it as a function of the institutional, market and macroeconomic variables assumed to have an impact on the speed of adjustment, denoted by the scalar:
Substituting equation (4) in equation (3) will result in the following model:
Working with a dynamic model such as (3) and (5) requires attention on the estimation procedure used as the lagged variable (is correlated with the error term and that is why the endogeneity problem arises. Applying an Ordinary Least Square model or fixed effects can give biased estimators: Huang and Ritter (2007) find that when applying the Ordinary Least Square method the estimators are upward biased that lead to a slower speed of adjustment and when using the fixed effects leads to a downward bias of the coefficient giving a much faster speed of adjustment.
One dynamic estimator that can eliminate this shortcoming is Generalized Method of Moments (GMM) that was suggested by Arellano and Bond (1991). The GMM estimator tries to remove the correlation between the transform error term and the lagged variable by using an instrument that is correlated with transformed lagged but not the error. Unlike the two mentioned approaches, the GMM estimator gives a consistent estimation by utilizing instruments that can be obtain from orthogonality conditions that exist between the lagged values and the disturbances: it reflects the parameter estimates so that the correlations between the instrument and the disturbances are as close to zero as possible as defined by a criterion function. [2]
资本结构的决定因素和速度的调整- Determinants of capital structure and speed of adjustment
A lot of studies have been conducted on the determinants of the capital structure. There have been documented that the capital structure varies across time and across firms because of the macroeconomic factors and firm's factors. Harris and Raviv (1991) determined that the leverage is positively influenced by fixed assets, non -debt tax shields, investment opportunities, and firm size and negatively by volatility, advertising expenditure, and the probability of bankruptcy, profitability and uniqueness of the product. Also, the studies of Rajan and Zingales(1995) and Bradley, Jarrell, and Kirn (1984) were conducted on the impact of the tangibility of assets (the ratio of fixed to total assets), the market-to-book ratio (usually thought of as a proxy of or investment opportunities), firm size, and profitability. |