人力資本、外溢效果與廠商生產力及技術效率：以台灣製造業為例

The Effect of Human Capital, Spillovers on Plant Productivity and Technical Efficiency：Evidence from Taiwan Manufacturing Industries

張景福(Ching-Fu Chang)

指導教授 ： 劉錦添

國立臺灣大學/社會科學院/經濟學研究所/博士(2010年)

https://doi.org/10.6342/NTU.2010.02054

摘要

本論文針對廠商探討三個與人力資本相關的研究議題。首先，在第二章中，我們從廠商的生產函數出發，分析勞工組成特性對廠商生產力與薪資的影響。運用獨特的台灣製造業雇主與員工合併資料，我們以本國員工的年齡、性別、教育程度之分布以及外籍勞工佔本國勞工比例做為衡量廠商人力資本的特性變數，並控制廠商的廠齡、規模與員工薪資分散程度等廠商特性。關於廠商生產力，我們則以生產附加價值、總要素生產力與Levinsohn-Petrin(2003)生產力作為衡量指標。研究結果顯示，若廠商規模越大、員工薪資分散程度越低、員工年齡組成比例越輕、男性員工比例與衡量廠內平均教育程度的人力資本指標越高者，廠商的生產力將會有所提升。而當廠商雇用外籍勞工時，其生產力會有所提升，但隨著雇用比例的上升將使貢獻度下降。至於平均薪資方面，各項特性對於薪資的影響方向大略與生產力相同，但員工年齡組成對於生產力與薪資則有相反效果。另外，外勞占本勞比率的上升分別對於技術勞工與非技術勞工的平均薪資有著正向與負向的影響效果，顯示外籍勞工與本國的技術勞工之間具有互補性，但與本國的非技術勞工則存在替代性。
其次，在第三章中，我們從廠商的生產技術效率角度切入，運用Battese and Coelli(1995)的隨機邊界模型以及豐富的員工特性變數，著手探討台灣製造業23個二欄位產業，人力資本對技術效率的影響。研究結果發現，全體製造業的勞動產出彈性約介於0.71~0.99之間，以皮革製品製造業最高；資本產出彈性約在0.13~0.36左右，以石油及煤製品製造業為首。至於平均規模報酬則以資訊電子業最高。各產業的平均技術效率估計值約介於0.61~0.78之間，以橡膠製造業居首位。另外，就技術效率影響因素方面，勞力或體力密集的產業(如民生工業、運輸工具及機械設備修配業)，僱用較高比例的年輕員工，將可增加其生產技術效率。而員工平均教育程度的提高有助於技術效率提升，但此助益將隨著員工平均教育程度的增加而遞減。除此之外，外籍勞工的僱用對所有產業廠商的技術效率皆有助益，顯示外籍勞工可填補一般國人較不願意從事的工作職缺，且增加國內勞工的相對競爭性。
最後，在第四章中，我們探討人力資本是否存在著外溢效果。然而，不同於既有的相關文獻，我們從廠商生產力變化來衡量該外溢效果的助益。運用各縣市除了該廠以外的製造業廠商員工之大學畢業比例作為外部人力資本衡量指標進行分析。然而，可能存在著許多未觀察到的因素同時影響廠商生產力與該縣市的人力資本，因此，我們藉由多種固定效果來排除不同年間、廠商與隨著時間變動的產業不可觀察異質性所可能造成的內生性問題。同時，我們也利用各縣市新開廠與關廠資料、1990年的人口普查資料以及勞動力調查資料創造出四個外部人力資本指標的工具變數進行估計。另外，我們亦執行多種穩健性檢測來確定人力資本外溢效果的存在性。實證結果顯示，當各縣市除了該廠以外的製造業廠商員工之大學畢業比例上升1%，將使得該廠的生產力上升0.93%~1.15%，且此外溢效果在高科技產業之間將更為顯著。另外，相關產業之間的外溢效果亦大於不相關產業之間的外溢效果，同時，該外溢效果隨著產業關聯性的減弱而逐漸消失。若將此外溢效果折算成金錢價值，則當各縣市除了該廠以外的製造業廠商員工之大學畢業比例上升1%時，該廠的生產附加價值將會提升約13,000元。

關鍵字

人力資本 ； 雇主與勞工合併資料 ； 勞工組成 ； 外籍勞工 ； 生產力 ； 技術效率 ； 外溢效果

並列摘要

This dissertation examines three empirical studies on the role of human capital in plant production. In the first study, we investigate the relationship between workforce composition and plant productivity among manufacturing industries in Taiwan. Based on an unique employer-employee matched data, we estimate production function using three alternative indices of plant productivity (including value added, Solow residuals, as well as Levinsohn and Petrin(2003) total factor productivity) and various workforce and plant characteristics (including the distribution of workers’ age, gender, educational level, the ratio of foreign workers, plant age, scale and wage dispersion). We find that larger plants, plants that employ a higher fraction of male, younger, higher educated workers and plants with less dispersion of wages perform better than others. Furthermore, hiring foreign workers is beneficial for plant productivity but the benefit decreases as the ratio of foreign workers rises. Finally, plants that employ a higher fraction of older workers would pay higher wages. Raising the hiring ratio of foreign workers is beneficial for the wage of skilled workers, but is harmful for the wages of unskilled workers. This indicates that there is a complementary relationship between foreign and skilled workers and a substitute relationship between foreign and unskilled workers.
In the second study, we apply the stochastic frontier model proposed by Battese and Coelli(1995) to investigate the relationship between human capital and technical efficiency among 23 two-digit manufacturing industries. Using an array of workforce characteristics such as measures of human capital, we find that the estimates of output elasticity with respect to labor are between 0.71 and 0.99, with the highest for leather industry, while the estimates of output elasticity respect to physical capital is between 0.13 and 0.36, with the highest for petrol and coal industry. Also, the coefficient of return to scale is largest for electronic industry. In efficiency analysis, the average technical efficiency for all industries is between 0.61 and 0.78. Further, employing a higher fraction of younger and high educational workers can improve the technical efficiency, but the former effect is only in labor or physical-strength intensive industries while the latter effect decreases as the average educational level of worker raises. Finally, hiring foreign workers is positively correlated with the technical efficiency for all industries, which implies that hiring foreign workers may stimulate native workers to work harder, and therefore, to be more efficient.
In the third study, we try to identify the existence and magnitude of human capital spillovers from production function. We measure the productivity gains as the college share of workers outside the plant within a city rises. However, there are some potential sources of endogeneity that may result in biased estimates. Consequently, we utilize multiple fixed effects as well as instrumental variables to cure the endogenous problem. Also, we carry out a series of specification and robustness checks. Our results show that one percentage increase in the college share outside the plant within a city will raise plant productivity by 0.93~1.15% and the effects are higher among high-tech plants and plants that belong to “Science Park” regions. Moreover, the spillovers between industries that are economically similar are stronger than spillovers between industries economically dissimilar. Finally, to assess the pecuniary benefits of spillovers, we find that one percentage increase in the college share of workers will increase the value added of each plant by about 13,000 NT dollars.

並列關鍵字

human capital ； employer-employee matched data ； worker composition ； foreign worker ； productivity ； technical efficiency ； spillover effect

參考文獻

江豐富 (2006), “外勞引進對本國勞工失業、職業選擇及薪資之影響”,《臺灣經濟預測與政策》，37(1), 69-111

林灼榮˙蔡榮德˙邱敬貿˙康家維 (2007) “研發投入與人力資本對台灣紡織業與通訊產業生產效率攸關性之研究”, 《東吳經濟商學學報》, 58, 89-120。

楊志海˙陳忠榮 (2001), “研究發展、專利與生產力-台灣製造業的實證研究”, 《經濟論文叢刊》, 30(1), 27-48。

蔡蕙安˙陳致綱 (2002), “研究發展外溢效果對廠商成本結構之影響-台灣高科技產業之實証分析”, 《經濟論文叢刊》, 30(2), 175-204。

歐陽利姝˙朱世琳 (2008), “研發、研發外溢與員工學歷結構差異對台灣資訊電子產業研發廠商的生產力之貢獻”, 《經濟論文叢刊》, 36(4), 515-550。