套利定价理论假设Arbitrage36800.pptx
Finance management Arbitrage Pricing Theory,APT Lecture by Hua-Ching,Cheng,PhD May,03,20042023/3/141Arbitrage Pricing Theory前言Forward套利定價理論假設Arbitrage Pricing Theory Assumptions 套利定價模式Arbitrage pricing model實證與測試Proof and test APTAPT 與 CAPM比較2023/3/142標準 CAPM 模式是基於平均值-變異數組合理論分析。投資者最好投資是基於選擇期望報酬與變異數 E(Ri)=Rf +(E(Rm)-E(Rf))Rolls critique(1977)CAPM中,市場投資組合(Market portfolio,E(Rm))是所有證券平均值與變異數效率組合。這種組合是事後效率(Ex post efficient),所以應該沒有證券 可以得到任何異常報酬(no securities have abnormal performance),但事實上證券組合存在異常報酬。因此所有證券所組成的市場投資組合不會是有效率組合 (Any Ex post inefficient index is possible)1.前 言2023/3/1432.APT假設Ross(1976,JET)發展套利定價理論 Arbitrage Pricing Theory(APT)。從套利的觀點來看資產價值訂定。APT 模式主要是基於套利單一價格法則(the law of one price):一種資產不會有兩種價格(two items that are the same cannot sell at different prices.)原來平均值變異數的架構替換成證券報酬產生的過程 (the process generating security returns.)APT模式,報酬產生過程是多指數模式(multi-index model)也就是說報酬率是由很多變數所影響,只是其中一項,還有很多因素會影響報酬率的形成的。2023/3/144APT假設同 CAPM 中所設定的均衡模式條件消費者也是相同的預期(homogenous expectations)。不同的是不必對投資者的效用函數作假設。2023/3/1453.APT模式說明基本型式(Factor analysis)Ri=E(Ri)bi1F1bi2F2bi3F3+bikFk iRi=第i個資產隨機報酬率E(Ri)=第i個資產期望隨機報酬率bik =第i個資產報酬率對第k個共同因素的敏感程度Fk=所有資產所面對的第k個共同因素i=第i個資產報酬率函數式的殘差2023/3/146Ri E(Ri)bi1F1bi2F2bi3F3iR1 E(R1)b11F1b12F2b13F31R2 E(R2)b21F1b22F2b23F32R3 E(R3)b31F1b32F2b33F33R4 E(R4)b41F1b42F2b43F34R5 E(R5)b51F1b52F2b53F35RpwiE(Ri)wii1Fiwii wi E(Ri)2023/3/147理論上,當殘差項為0,報酬可以為所有因子所解釋,截距 E(Ri)為預期平均報酬共同因子F的係數b,稱為因素負荷量factor weight or factor loading.共同因子F有下列關係存在,COV(Fi,Fj)=0 Cov(i,j)=0 Cov(Fi,i)=0 且 in(0,2)2023/3/1484.重要APT 條件wi0 在沒有套利的機會下,增加某一證券比重,就要減少其他證券持有的比例。套利不會增加投資者財富。Rpwi E(Ri)wibi1Fi wibikFk+wi iwi1/n,n,wibik0 在多個不同共同因素結構中,在沒有套利的機會下,不會因為操弄不同的因素,利用不同因素權重的比值,可以獲得額外的報酬。Rpwi E(Ri)wibi1Fi wibikFk wi E(Ri)wiE(R ik)0 表示在沒有套利的機會下,任何風險資產的組合,不會獲得額外的財富。2023/3/149Chen,Roll and Ross(1986,JoB)use macroeconomic variables such as.the growth rate in industrial production,unexpected inflation,the term structure of interest rates,and risk premiaBurmeister and McElroy(1988,JoF)also use macroeconomic variables such as a default risk,time premium,unexpected inflation,change in expected sales,etc.Sharpe(1982,JPM)uses finance/accounting type variables such as the return on the market(S&P 500),dividend yield,return on long-term bonds,firm size,etc.5.APT多指數實證研究2023/3/1410APT多指數影響因素工業生產指數成長率(IP,growth rate of industrial production)不同等級債券殖利率之差(risk premium):等級較差債券負擔較高風險溢酬,隱含市況不佳,投資人悲觀,對報酬率為負面影響。不同到期日債券利率之差(Term structure of interest rates):正常情況下,長期債券利率大於短期利率。石油危機,發生高通膨,兩者利率差加大,若沒有通膨壓力,與報酬率正相關不同預期的通貨膨脹率(Unexpected inflation)2023/3/14116.APT驗證基本上有兩種方法驗證 APT.第一種方法是使用因素分析法第二種方法是採用Fama and MacBeth(1973,JPE)兩階段步驟使用第一種方法如本章前所述。APT是一個報酬產生的多指數模式Ri=E(Ri)bi1F1bi2F2bi3F3+bikFk i2023/3/1412第二種方式,一樣先算出因素分析再利用因素負荷組成風險價格(the prices(s)of each risk)在均衡時,找出期望報酬在各個因素負荷下的最適組合Fama and MacBeth(1973)兩階段步驟2023/3/1413APT第二種方法說明使用 APT,將所有投資組合構成一個平面1 and 2 are the returns for bearing risks associated with indices(or factors).i are called factor prices(因素風險價格)(因素風險價格)and bij are called factor risks(因素風險)(因素風險).This is the equilibrium model produced by the APT when returns are generated by the two-index model.2023/3/1414 6.APT例題講解如果能夠完全多角化,僅會存在系統風險。影響系統風險,用bs因子負荷量表達.投資 人可用 APT多指數模型說明均衡狀況。為求簡化,設兩個指數模型。如下2023/3/1415例題討論2023/3/1416從上述三個組合,可以產生一個均衡平面的資訊可以用bi1,bi2因素負荷,作成一個平面將例題數據bi1,bi2帶入,求聯立方程式,可以得到下列公式與s.這一條就是APT模式下,均衡報酬率公式 2023/3/1417如前述,期望報酬與風險因子的關係下。2023/3/1418將 bs 帶入公式假設另有一組合 E,期望報酬 E(RE)=15%,bE1=0.6 與與 bE2=0.6.另一組合 D,其組合是由前面ABC三種證券所組合,各三分之一 D=A+B+C.使用前述報酬率公式,E(RD)=13%.組合D計算公式如下:比較市場上組合D與E2023/3/1419經由這樣的組合,投資者在組合 D 與 E,兩組風險是一樣的,都是b=0.6。風險一樣,但兩組的報酬率不同,組合E是15%,組合D卻只有13%,會產生套利機會。根據單一價格法則,兩個組合應該會有相同報酬,只有一個報酬。這時候,期初,套利者可以賣空組合D,比如說賣空(+100),同時買入組合 E,支出(-100).假設套利者的套利組合,符合前述零投資(zero investment),零系統風險(zero systematic risk),因此這個買進賣出,沒有增加投資額,期末,卻可以增加每100.有期望獲利2。套利者會持續套利到組合E 降到 A,B and C(and D)的組合平面上。也就是說組合E應該會降到13%報酬率。證明套利的過程2023/3/1420假設若存在下列三種情形組合 A.bp1=0bp2=0 B.bp1=1bp2=0 C.bp1=0bp2=1在組合A中,E(RA)=0 which is the return on a zero-b portfolio which we call RF(無風險利率)if a riskless asset exists.在組合B中,E(RB)=RF+1 which implies that 1=E(RB)-RF.在組合C中,E(RC)=RF+2 which implies that 2=E(RC)-RF.上述這些特殊情形表示,j 是異常報酬,異常報酬高低會受風險資產的影響。(only subject to risk j)套利特殊狀況說明2023/3/14217.APT與CAPM比較研究Chen,Roll and Ross(1986,JoB),Burmeister and McElroy(1988,JoF)and Sharpe(1982,JPM)all find that various factors are significant and their results suggest that CAPM can be rejected in favour of APT.While the APT model is more general the CAPM it does not say anything about the size or signs of the factor prices(s).The factors(Is)are defined for CAPM but not defined by APT.2023/3/14228.APT與CAPM公式證明The existence of the APT model is not necessarily inconsistent with CAPM.假設一個指數時。Suppose that assumed multi-index model for generating returns only depended on one index namely the return of the market portfolioThe APT equilibrium model isThis is the standard CAPM.2023/3/1423If the return generating process is more complex a similar result holds.Consider the two-index modelAs we have seen j is the excess return on a portfolio with a bij of unity on one index and zero on all other factors.The APT equilibrium model for this two-factor return generating process with a riskless asset is 假設兩個指數時假設兩個指數時2023/3/1424For certain portfolio j is an excess return.The second term in(a)is due to assuming that CAPM holds.Note if the js are not significantly different than i(E(RM)-RF)then finding two significant factors is fully consistent with CAPM.2023/3/1425Substitute(a)into(b)givesThus finding more than one significant factor in testing the APT model is not sufficient proof to reject any CAPM.This is the standard CAPM.2023/3/1426Maurice RocheMaurice RocheElmar MertensElmar Mertens(20022002)2023/3/1427The End2023/3/1428