生產力與技術效率之研究—台灣資訊服務業與農業之驗證

Abstract

本論文針對資訊服務業廠商與日治時期蔗作農場，探討生產力與技術效率的相關議題。首先，在第二章中，我們以2006與2011年之「工商及服務業普查」資料做為實證基礎，利用Seemingly Unrelated Regression模型檢驗影響資訊服務業無形投資的因素，接著利用Cobb-Douglas生產函數，探討台灣資訊服務業的無形投資對生產力之影響。本文將無形資產進一步分為：研究發展支出、技術購買、員工訓練支出、市場行銷支出與電腦軟體與資料庫購買等五個項目，檢視無形資本投資對生產力的整體效果與個別效果，同時探討各類無形資本對產出所造成的差異。實證結果發現：1.市場占有率、國外投資、資本密集度與廠齡對五種無形投資均有顯著影響。2.資訊服務業的無形資產具有提升生產力的效果，但貢獻度仍低於有形資產與勞動力。3. 廠商的研究發展經費、電腦軟體與資料庫購買以及市場行銷對生產力有顯著的正向效果，其中以研究發展經費的貢獻度最大；而技術購買、員工訓練兩變數對生產力並無顯著性影響。此外，透過穩健性檢驗（robustness check）分析的結果，發現以ROA做為獲利能力與績效指標時，無形投入對資訊服務廠商的生產力與獲利能力均有顯著的正向貢獻，同時，不同類型的無形資產對各行業廠商生產力的影響程度亦不相同。 其次，在第三章中，我們採用Battese and Coelli （1995）提出之隨機邊界模型，估計2000-2012年台灣資訊服務業上市櫃廠商的技術效率水準並探討影響技術無效率之因素。根據「台灣經濟新報（TEJ）資料庫」所提供之平衡縱橫資料，我們更進一步地考慮廠商的組織行為，從廠商特性變數與公司治理變數的角度切入，探討對技術效率的影響，進而做為改善廠商生產效率之依據。研究結果發現，台灣資訊服務業呈現固定規模報酬現象，Translog生產函數較適合作為研究期間之邊界生產函數；增加勞動、資本與中間投入的要素使用對提升廠商產量將有正向的幫助。在要素產出彈性方面，中間投入的貢獻度最大，接著依次為資本投入與勞動投入。而技術效率之估計值為0.845，故仍有提升之空間。就技術效率影響因子方面而言，R&D、廠商規模、營業內容、股權結構與董事會規模皆對技術效率產生顯著的正向影響或負面衝擊，因而也對研究結果所引申出之管理意涵進行了討論。 最後，於第四章中，我們以日治時期的甘蔗農業做為研究對象，同樣利用隨機邊界模型估計1926-1928年台灣蔗作農場的生產函數與技術效率，納入農場管理者以及環境特性，檢視效率之決定因素為何。實證結果發現，Translog生產函數較Cobb-Douglas函數型態更適合作為研究期間蔗作農場之邊界生產函數，投入要素則以肥料為影響生產的主要因素，而蔗作農場呈現固定規模報酬的現象，農場平均技術效率約為69％左右，並不具充分的技術效率。另外，技術無效率誤差項對甘蔗生產有顯著影響，在技術無效率模型所使用之變數中，經營者屬性、灌溉系統與自給勞動率等因素皆對技術效率造成影響，其中自耕農的技術效率高於佃農，而灌溉系統較方便的農場對效率之提升有所助益。

The main objective of this dissertation is to examine the level and the determinants of productivity and technical efficiency for information service industry and agriculture in Taiwan. In the first study, we investigate the impact of intangible capital on firm productivity, using a large panel dataset of Taiwan’s information service firms for 2006 and 2011. We also include various subcomponents of intangible capital (i.e. R&D, technology purchase, on-the-job training, marketing, software and database) into the production function all at once for identifying the relative contribution to the productivity of these firms made by different categories of intangibles during the study period. Our results indicate that the positive influence of aggregated investment in intangible capital on firm productivity level, but this contribution is lower than the contribution of labor and tangible capital. Moreover, disaggregating intangible items play different role on enhancing output and productivity, we also find significantly positive impacts on the productivity are R&D, marketing, and software and database, especially R&D. Finally, we evaluate further the relationship between intangibles and return on asset (ROA) to check the robustness of our results. We find that intangible investment can facilitate not only productivity, but also profitability. In the second study, we apply a stochastic frontier production model proposed by Battese and Coelli （1995） to estimate firm technical efficiency of Taiwan information service industry. In addition, we examine the determinants of technical efficiency and measure the effect of corporate governance mechanism on technical efficiency. We collect a balanced panel data of listed firms from Taiwan Economic Journal (TEJ) data set, covering the period 2000-2012. The empirical results suggest that managerial ownership and board size have a significant and positive correlation with technical efficiency. However, leverage（financial constraints）, executive compensation, larger shareholders and pledge fail to show significant impacts on technical efficiency. Moreover, for firms in our sample, R&D has a negative and significant effect on technical efficiency but increases firm size would help to improve the technical efficiency. The results show that Taiwan listed information service firms had been operating under constant returns to scale, and relied heavily on labor input and intermediate inputs. The study also provides evidence-based policy and management guidelines to promote technical efficiency and competitiveness of Taiwan information service industry. In the third study, we employ the stochastic frontier production model to estimate the technical efficiency of sugarcane production and attempt to examine the determinants of technical efficiency during the Japanese colonial period, from 1926 to 1928. The empirical results indicate that the sugarcane farms with constant return of scale fits better with the Translog function, and fertilizer is a major determinant in sugarcane production. The technical efficiencies for the farmers range from 0.21 to 0.94, with a mean of 0.69. Also, factors that make significant impacts on the production efficiency are land tenure systems, regional differences, irrigation systems, cultivation periods, cropping distance and the ratio of family labors to total labors.