【正文】 (2) Y = f(e, e)。從羅森（1990）的研究分析中可以找到所涉及問題的最佳解釋和發(fā)展的模型。研究分析過程中，我們發(fā)現(xiàn)了一些數(shù)據(jù)，表明在小企業(yè)中任職期間的CEO薪酬與其自身業(yè)績相關(guān)，CEO薪酬與公司業(yè)績敏感性較強。（經(jīng)濟學(xué)家報紙，1995年6月3日）在合理的市場環(huán)境下，薪酬的上升趨勢，可以通過調(diào)整勞動供給曲線來穩(wěn)定，也就是可通過左移或向上移動勞動供給曲線來實現(xiàn)。我們對人才內(nèi)部結(jié)構(gòu)做出假設(shè)：CEO薪酬是與控制在每個層次（s）和對層（不適用若干層跨度增加）參與，這樣就更能提高個人層次對這種結(jié)構(gòu)的影響。2 ? 0 (3)思考過程是這樣的：由于是自然狀態(tài)下的企業(yè)，所以受管理者個人業(yè)績以外因素作用的差異程度就顯得相對較小，這樣也就更利于體現(xiàn)個人能力對企業(yè)的影響力程度。 威斯巴赫 1988）。 that is, managers must have bee more productive. Managers can increase their productivity either by controlling more resources or by controlling a constant amount of resources more effectively. We here address the following question: Does the market for top executive work? In order to provide an answer to this we test the twin predictions of the agency model that (i) executive pensation and economic performance are positively correlated and (ii) that poor pany performance results in a higher probability of CEO turnover. The former proposition has been tested for both the . and the . using, in general, data for large quoted panies. A recent review of this literature is presented in Conyon, Gregg and Machin (1995).This paper augments the . empirical literature on the determination of top pay by focusing on executive wage setting in small panies。vi, t – 1 represents some measure of performance (. profits per employee, accounting rates of return, shareholder rate of return, sales, firm profits relative to industry profits)。1 0, f162。詹森和墨菲（1990）有影響的論文中也報告說，薪酬與績效之間的敏感度低，股東財富每增加1000美元CEO的財富增加 。然而，可以通過其他手段來規(guī)范管理行為。 f 162。2 分析框架上世紀(jì)六十年代開始關(guān)注對高管薪酬的理論研究，但是隨著信息經(jīng)濟的發(fā)展，它只持續(xù)了一段時間。在這類型的企業(yè)管理中首席執(zhí)行官的6位數(shù)薪酬遠(yuǎn)遠(yuǎn)低于大型上市公司，但即使如此也比我們預(yù)期的要高很多。因為由于管理培訓(xùn)和商學(xué)院的課程擴散，供應(yīng)給社會的合格管理人員人數(shù)就可能得以增加，勞動供給曲線就有可能轉(zhuǎn)移到了右側(cè)，因此，在一定范圍內(nèi)至少有一個管理人員愿意增加工資的數(shù)量來做這項工作；而對于上移需求曲線時，管理人員的邊際產(chǎn)品必須增加，也就是說，管理者必須要更加富有成效。因此，所期望的是有更多的管理信息是反映高管薪酬與企業(yè)規(guī)模大小是呈正相關(guān)的。但是這個問題比較復(fù)雜，因為即使某一個行業(yè)可能受到宏觀經(jīng)濟環(huán)境影響程度類似，但接觸不利沖擊的程度的大小卻是一個由不同管理決定而導(dǎo)致。一個基本的模式是如下的形式： Pr(DCEOit) = F(xb) =exp(xb)/(1 + exp xb) (5)其中：CEOit表示一位首席執(zhí)行官以新的或其他方式存在；在x向量中包括了在過去的P時期公司的平均表現(xiàn)； 加權(quán)是表示首席執(zhí)行官的特點，qkt,一些不隨時間變化的（如出生日期， 國籍）和有些是隨時間變化（如選擇任用，教育）的量。 an area that has received paratively little attention (an exception is Storey, Watson and Wynarczyk, 1995). The second innovation herein is the focus on CEO turnover. The influential paper by Jensen and Murphy (1990), based on data for large . firms, illustrates that the discipline provided by the threat of job dismissal acts to align the objectives of managers and owners. They find that management turnover is greater in panies with poor economic performance as predicted by the agency theory. There is, however, a paucity of published evidence on the executive turnover mechanism in . panies. Furthermore, the studies that do exist relate to large rather than small firms (see Cosh and Hughes, 1995).The current paper hopes to add to the understanding of managerial labour markets by assessing whether managers in small firms are disciplined for poor pany performance by increased probability of turnover.Th