| 摘要: | 研發活動於促進經濟成長與維繫永續成長上，扮演不可忽略之地位。各國政府均視推動研發活動之發展為首要目標。創新科技政策的著力點，除積極誘發更多資源投入外，亦可思考如何將有限的研發資源做妥善的配置，以強化研發活動之成效。目前，有些跨國組織之報告已意識到，OECD各國研發資源，在不同技術密集產業之間的配置，已有所差異，研發資源有集中於特定產業或技術領域之趨勢。基於上述的背景，本論文之目的將以一國『研發資源跨產業間的配置結構』為研究主軸，探究三個相關之議題。研究結果分別呈述如下： 議題一，鎖定於探究影響研發資源配置的因素。分析之觀點著重於一個國家整體研發資源之佈局考量。鎖定之焦點有兩個方面，一是檢視產業的未來需求面力量與過去創新經驗，是否牽動著一國的研發資源佈局。另一方面則是探究，驅動國家研發資源配置的力量是否與多國籍企業的海外研發佈局有關。本議題以17個OECD國家作為研究對象，資料期間涵蓋1982至2000年。因考量到研發活動投入與其配置之間的關係，本議題採用聯立方組，以三階段最小平方法(3SLS)進行估計。實證結果發現，產業內的市場需求之量的確能夠驅動資源投入於該產業，驅使資源呈現較集中之現象。但是，產業內過去創新經驗對於研發資源的吸引力效果卻不明顯，原因可能來自於各國錨定之新興產業政策有關，進而影響了技術拉力對於資源配置的影響。另外，實證結果亦支持多國籍企業的海外佈局會對研發資源配置的集中趨勢有正面影響。 議題二，採用追蹤資料的Granger因果檢定，論證一國研發資源配置態勢與其所得不均之間的關係。一國研發活動之配置主要係關注其研發資源投入在哪些產業。換言之，研發配置的形態可用分散或集中於某些產業來衡量。本議題檢視之問題有二，一是不同研發配置結構是否會影響所得不均。在考量研發分配之角色時，若一國的研發資源集中且偏向於特定產業，可能造成資源（如勞動與資本）有跨產業流動之現象，進而驅使所得不均現象之產生。另一問題則著重在反向關係，因為具所得不均現象的社會，可透過需求拉力或是人力資本累積機制，影響研發活動的配置態勢。本論文分別採用不同衡量研發配置指標、實證模型設定與資料（研發支出與專利）來檢視這個問題。以6個OECD國家（加拿大、芬蘭、英國、挪威、瑞典與美國）作為研究對象，資料期間涵蓋1987至2000年。實證結果發現，跨產業研發配置的改變對所得不均存在Granger的因果關係。反向因果關係之證據，則相對較微弱。 最後，議題三則聚焦於研發資源分配與生產力成長之關係。主要的問題在於什麼樣的國家研發政策有助於生產力成長？集中於少數產業，還是分散於較廣泛的產業呢？本議題以11個OECD國家作為研究對象，資料期間涵蓋1981至2000年。因考量國家本身之異質性、跨國之間的相依性與序列相關的問題，本議題採用固定效果且考量橫剖面相依性（cross-sectional dependence）之追蹤資料模型進行估計。此外，考量到研發活動性質之差異，本議題將研發活動分配區分為研究型（research-oriented）與應用型（development-oriented）配置。實證結果發現，集中的研發配置對生產力成長有正面的影響，但是集中程度太高時，反而會降低正面影響效益。另外，研發密集對生產力成長影響之正面效益，會隨著研發資源集中程度而改變。由上可知，適當將研發資源集中於少數特定產業內，能有助於生產力成長。此外，在分群樣本方面，實證結果支持集中資源之策略施行於發展型產業，對提昇生產力成長有較大之效益。而於研究型產業之效果則較低。 There is a growing awareness tendency concerning the differences in the allotment of R&D resources to different industries in Organization for Economic Cooperation and Development (OECD) countries. The phenomenon of more and more R&D investment being concentrated in only few industries has been noted. In addition to promoting the increases in the overall R&D activities, drawing up a scheme for the appropriate and effective allocation of these resources may be regarded as to a useful tool for propelling the development of R&D activities nationwide. With this background in view, the objective of the dissertation concerns three issues relating to allocation of R&D resources. In first issue, this paper is to investigate on two aspects affecting the allocation of R&D resources across industries, that is, market-pull and technology-push forces of industry and internationalization. By using a panel data on 17 OECD countries over the period 1982-2000, this paper constructs a recursive simultaneous equation model and then estimate by three-stage least squares (3SLS). The findings support that R&D decisions with the consideration of targeting unmet needs of markets can drive to increase R&D inequality. However, the effect sourced from technology-push force on allocation of R&D investment is relative weak. There is a possibility that some innovation policy aiming at the development of newly emerging or niches industries in R&D activities is a cause of this result. In addition, the empirical evidence indicates that MNCs’ oversea R&D activities can influence the extent of R&D specialization in certain industries in host countries. In the second issue, this study applies a panel Granger causality test in heterogonous panel data models to investigate the relationship between R&D inequality and income distribution in six OECD countries (Canada, Finland, the United Kingdom, Norway, Sweden, and the United States) over the period 1987-2000. We are paying more attention to the role of the composition of R&D activity within a country so as to reflect the differences in R&D inequality across countries. In addition, this study also checks the robustness of the results by using different measurements for R&D inequality, the lag-order of the specification model, and datasets. The empirical results found that changes in the inequalities of R&D resources across industries in a country Granger cause changes in income inequality within that country. The results are found to be robust, and give support to the widely suspected, still ongoing trend in recent years that the advent of the knowledge economy in the form of greater R&D spending may contribute to rising income inequalities if such spending is concentrated in a few industries. It is also found that the reverse direction is relatively weak and not robust. Finally, this study is intended to underline the important role played by the cross-industry allocation of R&D resources, as indicated by several indices measuring R&D inequality, within the manufacturing industries of each country, and how it determines cross-country productivity growth. In other words, the main issue is whether national R&D policies for promoting productivity growth should be narrowly focused or spread across a wide range of industries. The allocation modes of R&D resources are divided according whether they are research-oriented or development-oriented in type. This is done to highlight the role played by differences in the properties of R&D activities for a particular industry. In the estimation process, the phenomena of group-wise heterogeneities, serial correlations, and the correlation between pair-wise country comparisons in the sample are taken into account. The fixed-effect panel data model with cross-sectional dependence is employed on panel data drawn from 11 OECD countries for the period from 1981-2000. When the other variables are controlled, we find that the effect of R&D inequality on productivity growth is significantly positive and the coefficient of its squared term is negative. Moreover, the impact of R&D intensity depends on the degree of inequality. This lends support to the notion that concentrating R&D spending on a few suitable industries is an important factor in encouraging cross-country productivity growth. Specialization of R&D activities in certain relevant industries creates better opportunities for promoting productivity performance, because of reaching sufficient levels of scale and allowing for accumulation of experience. Furthermore, the results reflect the fact that development-oriented manufacturing industries enjoy higher productivity growth as a result of this specification policy. However, the case for research-oriented industries is relatively weak. |
