廠商規模與存活：受限制分量迴歸之應用

Firm Size and Survival Duration - An Application of Censored Quantile Regression

嚴婉禎(Wan-Chen Yen)

指導教授 ： 朱珊瑩

中原大學/商學院/國際貿易研究所/碩士(2007年)

https://doi.org/10.6840/cycu200700746

摘要

過去在產業面的文獻中，對於廠商規模與存活的關係研究，我們發現初始規模與存活存在不同的結論，顯示兩者之間為無關、正相關及負相關。另外，如同本研究，在以上市公司為觀察對象的文獻中，也發現公司規模與存活之間的關係其結論不一致。Geroski (1995) 認為廠商通常以小規模參進，且預期存活期較短。而多數實證結果也指出規模小的廠商其存活期也較短。此外，Jovanovic (1982) 指出有效率的廠商透過對環境學習及調適進而成長及存活，但此調整及學習是需要時間。因此，本研究認為在存活期間短，經驗效果尚無法發揮，此時初始規模對存活的影響較為重要；然而當存活期間較長時，廠商擁有充足的經驗，規模的重要性相對不那麼重要。我們以台灣在1988年1月到2007年4月電子業上市公司作為研究對象，以股票發行規模作為規模的衡量，利用受限制分量迴歸 (CQR) 作為分析工具，觀察初始規模對存活的影響效果，是否隨存活期的長短不同而不同，為先前結論不一致的情況作一協調。並與Cox Proportion Hazard Model (CPH) 結果作比較，指出傳統模型的缺失。結果顯示對存活期大約為3到5年公司而言，初始規模對於存活是具有影響力，但在高分量，即存活期相對較長者，影響較不顯著。並且對於存活期較短者，初始規模對於存活具有正向影響。除了規模，在規模平方項、上市前的成立數及市場活絡度的CQR結果中，皆顯示出變數的影響效果也隨分量的變化而變化。但CPH的結果皆為不顯著，表示在平均值條件下分析結果被平均化，並無法看出初始規模效果的變化。

關鍵字

存活 ； 受限制分量迴歸 ； 存活期 ； 廠商規模

並列摘要

The relationship between firm initial size and survival is different within some industrial studies. Some studies indicate that they are not related, some studies reveal that they are positively or negatively correlated. In addition, some empirical studies analyze the relationship between performance and firm size of listed firm. The outcomes are also discordant. We think the controversy is the result of lack of duration effect. Firms usually enter with small size and we usually expect survival duration of them is short (Geroski, 1995). Furthermore, a lot of empirical works find that small firms usually have shorter survival duration. Besides, Jovanovic (1982) conclude that efficient firms would grow and survive, but it would take time to find their efficiency. However, if survival duration is not long enough, the effect of experience will not work. Therefore we infer that, in the short survival duration, size does matter. But in the lung survival duration, because of the effect of experience or learning effect, size is not that important.
Due to the limitation in data availability, we use listed firms of electronic industry in Taiwan, from January in 1988 to April in 2007. We use offering stock volume as size proxy. We imply censored quantile regression to observe whether the relationship between firm size and survival duration change, and compare the outcome with Cox proportional hazard model.
The results from CQR indicate that the effects of firm size are changed. Firm size and survival are correlated in the law quantiles, but are uncorrelated in the high quantiles. As far as firms with long survival duration concerned size does not affect. However, firms who live short, firm size is concerned. In the short duration, they are positively correlated. The result is consistent with many findings. Large firms are easy to survive. In addition to size, results in square term of size, age at offering and offering activity reveals effects are changed in different quantils. However, the results from CPH are not significant, because CPH is in the mean conditional. If we use CPH as analysis tool, the real effect may be averaged and we could not see the changes of size effect.

並列關鍵字

firm size ； censored quantile regression ； survival duration ； survival

參考文獻

江志民、李勝輝、李宜穗、周宇超 (2007)，「技術創新活動典電子業的影響」，《數據分析》，2 (1)，61-73

Tsai, S. L., and C. M. Kuan (2004), A Quantile Regression Analysis in Family Income Determination with Mating Effects, Working Paper, Institute of Sociology, Academia Sinica.

李春旺、劉維琪、高孔廉 (1989)，「股票行為與規模效應：台灣股票市場實證研究」，管捚評論。

莊碩立 (2004)，「台灣汽車車體險理賠因素分析：受限制分量迴歸之應用」，國立台灣大學財務金融研究所碩士論文。

Guimaraes, J., J. A. F. Machado, and P. Portugal (2004), Has Long Become Longer or Short Become Shorter? Evidence from a Censored Quantile Regression Analysis of the Changes in the Distribution of U.S. Unemployment Duration, Unpublished discussion paper, Universidade Nova de Lisboa.