由於研究發展投資對企業價值有正向的影響;而在我國目前稅制下,股利政策對企業價值似乎存在極微弱之效果。本文將不同研究發展投資抵減稅率上限為例將期間分為三個階段(87及88年、89及90年、91至93年),其目的是想從不同投資抵減上限之期間中探討企業研究發展決策、股利政策及公司價值之因果關係。 本文利用SEM模式的各項指標( 值比率、RMSEA、GFI指數、IFI指數、NFI指數、CFI指數、RMR指數等項目)來評估模型契合度。在原始模型中,87、88年此階段有四個路徑參數未達顯著性,89、90年此階段有更有高達五個路徑參數未達顯著性,根據原始模型的殘差分析指出「資本密集度」、「負債比率」、「融資政策」、「公司前期盈餘成長」均與「企業價值」的測量殘差之間具有直接關係;因此本文將87、88、89、90年之所有樣本作為一個結構模型,重新再進行一LISREL模型配適度及效果分析,並繪製路徑圖。而87、88年及89、90年這兩階段中,其模型所呈現出的結果較為一致,故本文在做模型修飾時,將此兩階段合併為一個階段做修飾。 在經過模型修飾後,模型契合度各指標皆達理想狀態,且各路徑參數的t值亦大多達顯著水準;各變項間的路徑參數結果,在直接效果、間接效果與整體效果的結果亦十分理想。根據本研究的假設模型,除了各變項之間的直接效果之外,銷售額、國民生產毛額、產業研究發展密集度、現金流量率、負債比率、融資政策均會透過研究發展投資對企業價值產生間接效果;公司規模亦會透過股利政策對企業價值產生間接效果。 本文研究發現,整個PA-OV路徑模型中,研究發展投資與企業股利政策對企業價值所影響的參數逐年減低(尤其以股利政策對企業價值之直接影響效果最為嚴重);反觀其他外衍變項(例如資本密集度、負債比率、融資政策)對企業價值的直接影響效果較前述兩者大;透過中介變項而對企業價值產生間接效果的外衍變項所產生的間接效果,亦比前述兩者之直接效果來得大。 關鍵字:公司價值、研究發展、股利政策、結構方程模式、LISREL
As the investment on research and development has a positive effect on corporate value, under the current tax system in Taiwan, the dividend policy seems to have an extremely weak effect on corporate value. This study used the ceiling of deduction tax rate for research and development investment as an example and divided the sampling period into 3 stages (1998-1999, 2000-2001, 2003-2004) for the purpose of investigating the causation of research and development decisions, dividend policy, and corporate value in the periods of different investment deduction ceilings. This study adopted the various indices of Structural Equation Model (SEM) ( ratio, RMSEA, GFI index, IFI index, NFI index, CFI index, RMR index, and etc.) to evaluate the fitness of the evaluation model. In the original model, the period of the year 1998-1999 had 4 path parameters under significance level, and that of the year 2000-2001 had as many has 5 path parameters under significance level. According to the residual analysis of the original model, it was shown that “capital intensity”, “debt ratio”, “financing policy”, and “growth of earning in the previous period”, are directly correlated with the measurement residual of "corporate value”. Thus, this paper created a structural model with all the samples from the years of 1998, 1999, 2000, and 2001, and further conducted a fitness and effectiveness analysis of LISREL model. The path diagram was also drawn. In the stages of 1998-1999 and 2000-2001, the results presented by the model were more consistent. So in modifying the model, the two stages were combined as a stage for modification. Through model modification, the fitness and various indices of the model were ideal, and most of the t values of each path parameter reached the significance level. The path parameters of each variable produced ideal results in the aspects of the direct effect, indirect effect, and overall effect. According to the hypothesized model proposed in this study, in addition to the direct effects of variables, sales revenue, GNP, intensity of research and development in industry, cash flow ratio, debt ratio, and financing policy will cause indirect effect through the investment of research and development on corporate value. The company scale will also create indirect effects through the dividend policy on corporate value. It was discovered that in the entire PA-OV path model, the impact of investment of research and development and the corporate dividend policy on corporate value was on the decrease year by year (especially the direct effect of the dividend policy on corporate value). On the contrary, the impact of exogenous variables (such as capital intensity, debt ratio, and financing policy) on corporate value was greater than the afore-mentioned two impacts. The indirect effect caused by exogenous variables through intervening variables on corporate value was also greater than the direct effect of the afore-mentioned two impacts. Keywords: Corporate value, Research & development, Dividend policy, Structural Equation Model, LISREL