This dissertation is about research and development (R&D) and formation of research joint ventures (RJV). The first chapter analyzes R&D competition and cooperation regimes with coordination costs under full information sharing and no spillovers in a Stackelberg model. The findings show that profits and R&D incentives of RJV members decrease with coordination costs. R&D cooperation of leaders results in higher profits for insiders and also higher welfare compared with R&D competition. Checking the robustness of the results shows that with R&D spillovers and no information sharing no RJV forms. With convex costs, an RJV containing all leaders forms. The second chapter considers a duopoly Cournot model where production may result in environmental damage. Firms can either invest in process or environmental R&D. In the first case, we assume an exogenous emission tax. With high enough emission tax, welfare is always higher under public R&D than cooperation. Under endogenous emission tax, when the regulator acts before firms’ decision on R&D, with high R&D spillovers, public R&D yields higher welfare than R&D cooperation. When the regulator sets the emission tax after firms’ decision on R&D, welfare under R&D cooperation is higher than public R&D. Comparison of commitment and no commitment also shows that commitment increases private R&D. Chapter three investigates the endogenous formation of coalitions under the size announcement game in a Cournot framework and analyzes the effect of coordination costs on equilibrium and optimal coalitions. When there are industry-wide R&D spillovers numerical simulations show that with high enough coordination costs no RJV forms in equilibrium, which also maximizes welfare. When there is intra coalition full information sharing and no inter-coalitions R&D spillovers with high enough coordination costs, the equilibrium coalition structure is more concentrated than when coordination costs are low and the size is higher than when RJVs could not form endogenously. Also, with high enough coordination costs, the welfare maximizing coalition is less concentrated than the equilibrium one while the opposite is true for low coordination costs.

經濟部技術處長久以來為推動產業研究發展工作、強化產業之技術開發能力，每年均委託法人機構執行「科技研究發展專案計畫」（科專計畫），對國家科技水準及人才培育，貢獻斐然，以為我國科技研發能力奠定良好基礎。為能對民間所具有研發能力之廠商加以整合應用，以利政府達成協助國內產業升級轉型，以提升國家整體競爭力的目的。自86年度起，經濟部科專計畫以開放給民營事業申請，一方面希望能夠借重民間企業的研發能力與資源優勢，提升原財團法人較不具執行優勢之科專計畫研究成果品質。另一方面藉由政府與民間企業研發單位之間的合作，亦可藉由產業界本身對科技研究之投入與努力，達到政府藏技於民的美意。其計畫目的以導引企業投入研究發展工作，並期以發揮相乘效果，促進企業進一步參與及進行投資，達到強化研發能量之目的。由於86、87試辦年度廠商反應熱烈，為擴大業界參與科技專案計畫，自88年度起，新作法改為以補助款方式來鼓勵業界投入研發活動（原為委辦方式），推動「業界開發產業技術計畫」（原業界科專）。且放寬申請資格、及智慧財產權下授給廠商…等變革。 由於88以後年度業界科專制度設計的改變，為瞭解兩套制度設計如何影響廠商參加業界科專的動機，本文應用理論模型、實證、深入訪談的方式來探討，得到若干結論如下： 1.由本文理論模型分析可知，86、87委辦年度時，從模型中推論可知，廠商申請業界科專為研發投入規模較大時，個別或會員廠商參加業界科專的意願較高。而88以後年度時，廠商參加業界科專後，由模型推論為不論此專案的研發規模為何，個別或會員廠商皆願意參加業界科專，這也符合經濟部擴大申請範圍的立意，而實證分析結果也大致上符合模型的推論。 2.委辦年度所設計的機制，較能鼓勵研發投入龐大及規模制度良好的廠商參加業界科專。若廠商參加業界科專研發的主要目的是希望政府共同分攤高風險、金額龐大的研發成本，或其研發成果較偏向基礎研發，或原先就具相當好的研發體制，在市場已佔一席之地而希望取得技術領先的地位者，在88以後年度的設計下，反而會降低其來申請業界科專之動機。 由於88以後年度的設計，下放智產權給廠商、提高廠商配合款比例及簡化申請手續。若參加業界科專主要目的在於以研發成果（應用研發）來獲取商業利益，及完全擁有智產權，且參加科專不但能與政府一起分攤研發成本、更能改善廠商體制，故較能吸引較新成立或規模較小的廠商來參與。 3.為瞭解廠商為何很少以共同研發的方式來參加業界科專，本文輔以深入訪談廠商的方式來瞭解。在委辦及補助年度各選取一家廠商來訪問。台中精機因研發成果技術上的需要，需與跨領域的錸德公司合作。由於此共同研發為獨立作業，除了在評鑑時台中精機有等待錸德外，共同研發並無太大問題。其參加主要的目的為研發經費的補助，並認為由於為私人企業，老闆較重視短期的利益，且研發又具高度風險，經濟部能給予廠商一筆經費來從事研發，老闆較不會限制其經費的運用，對業界研發助益很有幫助。但由於廠商自籌款的要求增加，並無再參加業界科專的動機。 另漢翔航空工業則表示，其研發聯盟伙伴皆為其下游廠商，由於此專案為建立一標準化作業系統，也因為如此，對其研發伙伴的包容性較一般研發聯盟還大。因其為領導廠商，相對的付出成本也更多。漢翔也是希望政府的補助款能再增加，否則對於領導廠商而言，參加業界科專成本反而更增加了！

The main goal of the diploma thesis is to find out how the import substeps from China participate in making a price. The import substeps of products classified within the section 7 of the Standard international trade classification are analyzed. Another goal of the thesis is to discover advantages and disadvantages of JV co-operation established in China. The results of partial analyses become the basis of making guidelines for those, who are planning to establish this kind of co-operation in chinese region.