Arbitrage Pricing Theory

(APT):
 Arbitrage Pricing Theory (APT):

Assumptions:
1. Capital markets are perfectly competitive.
2. Investors are rational.
3. The stochastic process generating asset returns can be
expressed as linear function of a set of K risk factors (or indexes)

In contrast to the CAPM, the APT does not assume

1. That investors can freely borrow and lend at the risk-free rate
2. That investors select portfolios on the basis of the mean and
variance of returns.
3. A market portfolio that contains all risky assets.
 Arbitrage Pricing Theory (APT):

• According to APT, returns on an individual stock in, say,

coming year will depend upon a variety of anticipated
and unanticipated changes in the economy over that
period of time.
• Changes in the Economy
• Rate of Inflation;
• Rate of Interest Rate
• Rate of Industrial Production
• Liquidity of the economy
• GDP
• Thus, there is a systematic movement of return because
of these changes.
 Idiosyncratic Factors
• Assets returns are also affected by factors that
are not systematic to the economy as a whole.
• We call these forces idiosyncratic to distinguish
them from the systematic forces.
• Assume that there are n underlying factors generating returns
for a security.

• The following equation illustrates the return generating process

of stock i:

• Ri = αi + βi1I1 + βi2I2 + ….+ βinIn + ei ….(1)

• where;
• αi = Expected level of return of the stocks when all indices have a value of zero.
• Ij = Indices of underlying factors
• βij= Sensitivity coefficients with respect to the underlying factors
• ei = Idiosyncratic Risk
• E(e) = 0 and E(eiej) = 0 as is true with single index model.
• Similar to the CAPM model, the APT
assumes that
• the Idiosyncratic factors (ei) will be
diversified away in a large
portfolio.
• For equilibrium to exist, the following
conditions must be satisfied:
• Using no additional funds and without increasing
risk, it should not be possible to create a
portfolio to increase return.
• S. Ross derived the following equilibrium
relationship which is referred to as the APT
model:
• E(Ri ) = Rz + i11 + i22 + ………+ ikk
 COMPARING THE CAPM AND THE APT

CAPM APT
Form of equation Linear Linear
Number of risk factors 1 K(≥1)
Factor Risk Premium [E(rm)-RF] λj
“Zero-Beta” return RF RZ
• According to the APT there are many factors that affect
returns, in contrast to the CAPM, where the only relevant
risk to measure is market beta.
• However, when we apply the theory, the factors are not
identified.
• In an empirical study the factors can be identified.
• There may be three, four, or five factors that affect
security returns.
 APT Model
S. Ross uses an arbitrage arguments to develop a model of
equilibrium pricing.

• Assume there is a single factor so that Ri = α + iI1 + ei

• and the equilibrium risk-return relationship, derived from the APT.

• E(Ri) = RZ + 1 

• Where
• RZ = A risk free return or zero beta return
•  = The risk premium
• 1 = Responsiveness of the security to changes in the single factor, i.e., measure
of the systematic risk.
 Arbitrage Pricing Model: Single factor

• E(r )

• RU = 15.0 U

• RA = 13.4 A
• C
• RB = 10.6 B

• βB=.8 βU=1 βA=1.2

 Process of Arbitrage:

• Portfolio U represents a profit opportunity, which can be converted

into a riskless arbitrage.
• Construct portfolio C comprising 50% investment in portfolio A and
50% investment in portfolio B.
• Therefore,
• C = ∑Wii = (0.5)(0.8) + (0.5)(1.2) = 1
• RC = ∑WiRi = (0.5)(10.6) + (0.5)(13.4) = 12%

• For Portfolio U:
•  U = C = 1
 Riskless Arbitrage

Investment return Risk

Beginning End (Beta)
Long Position of Portfolio U -Rs1,000 Rs1150 Rs150 1.0

Short Position of Portfolio C Rs1,000 -Rs1120 -Rs120 -1.0

Arbitrage Portfolio 0 Rs30 Rs30 0.00
 APT risk-return relationship
• Suppose stock return is generated by a single
factor model.
• What is the relationship between the expected
rate of return and the responsiveness β1 to the
factor?
• Suppose the relationship is nonlinear as shown
in the following figure.
• Then, any one can make unlimited sums of
money with no required investment and no risk.
 Infeasible relationship between E(r )
and β1,j
in one factor model

• E(rj )

• F
• E
• D

• C
• B
•
• A

• Factor Beta β1,j

 Infeasible relationship between E(r )
and β1,j
in one factor model

• E(rj )

• F
• E
• D

• C
• B
• E(rZ′ )
• A

• Factor Beta β1,j

 Infeasible relationship between E(r )
and β1,j
in one factor model

• E(rj )

• F
• E
• D

• C
• B
• E(rZ′ )
• A

• Factor Beta β1,j

 Infeasible relationship between E(r )
and β1,j
in one factor model

• E(rj )

• F
• E
• D

• C
• B
• E(rZ′ )
• A

• E(rZ )

• Factor Beta β1,j

 Infeasible relationship between E(r )
and β1,j
in one factor model

• E(rj )

• F
• E
• D

• C
• B
• E(rZ′ )
• A

• E(rZ )

• Factor Beta β1,j

• There are an unlimited number of securities
lying on the curved line of the above figure. Six
of the securities are labeled at points A, B, C, D,
E, and F.
• Combination line for stocks C and E is given by
the line passing through points E(rZ′ ), C, and E.
• Positions between C and E are taken by
investing positive amounts of money in both
stocks.
• Positions between E(rZ′ ) and C are taken by
selling stock E short and using the proceeds to
invest in stock C.
• Note that by selling stock E short and investing
in C, we can construct a portfolio positioned on
the graph at point E(rZ′ ).
• The Beta of this portfolio is equal to zero. We
have constructed the portfolio at E(rZ′ ) by using
two stocks, C and E, but we could have also
constructed it by using four, shorting stocks E
and F and using the proceeds to invest in stocks
C and B.
• In fact we could have constructed the same
portfolio by using as many pairs of stocks as we
want.
• The portfolio has no systematic risk and
almost no residual variance, but it has an
expected (riskless) rate of return equal to
E(rZ′ ).
• Note that we can construct another
portfolio positioned at E(rZ ) by selling
stock D short and investing in stock A.
• Again employing an infinite number of
pairs of stocks, we can construct a
portfolio positioned at E(rZ ) with virtually
no systematic risk or residual variance.
• We have now constructed two zero-variance, or
riskless, portfolios with two different expected
rates of return.
• Now we can sell a given amount of portfolio
positioned at E(rZ ) short and using the proceeds
(with no equity investment of our own) to invest
in the portfolio positioned at E(rZ′ ).Assuming
there are no restrictions on short sales, we can
earn huge profit.
• Everyone will try to take this opportunity, selling
short stocks such as D, E, and F while buying
stocks such as A,B, and C.
• We will drive down the prices of stocks such as
D, E, and F and drive up their expected rates of
return.
• In the same sense our buying activity will drive
up the prices of stocks such as A,B, and C and
drive down their expected rates of return.
• The effect of all this will make the relationship
between expected return and factor risk linear,
as in the following figure.
• Thus, in a single-factor APT, the relationship
between factor risk and expected rates of return
is given by
• E(ri) = RZ+ 1 .
 A Feasible relationship between E(r )
and β1,j
in one factor model

• E(r )
•
• F
• E
• D
• C
• B
• A
• RZ

• Factor Beta β1,j

Imagine that an investor owns three stocks and the current market
value of his or her holding in each stock is Rs.4,00,0000.
Total investment W0 = 12,000,000. Everyone believes that these three
stocks have the following expected return and sensitivities:

Stocks ri bi
1 15% 0.9
2 21 3.0
3 12 1.8

• Do these expected returns and factors sensitivities represent an

equilibrium situation?
• If not, what will happen to stock prices and expected returns to
restore equilibrium?
 Arbitrage Portfolios
• If Xi denotes the change in the investor’s holdings of security i (and
hence the weight of security i in the arbitrage portfolio), this
requirement of an arbitrage portfolio can be written as:

• X1 + X2 + X3 = 0 …………. (2)

• Second, an arbitrage portfolio has no sensitivity to any factor:

• b1X1 + b2X2 + b3X3 = 0 ………….. (3)

• Or in the current example:

• 0.9X1 + 3X2 + 1.8X3 = 0 …………. (4)
• Let, X1 = 0.1
• Therefore, 0.1 + X2 + X3 = 0 ………….. (5)
• 0.09 + 3X2 + 1.8X3 = 0 ……………… (6)
• Solving these, we get, X2 = 0.075 and X3 = -0.175. Hence, a
potential arbitrage portfolio is one with these weights.

• The third and last requirement for an arbitrage portfolio is:

• X1r1 + X2r2 + X3r3 > 0 …………(7)
• Or for this example,
• 15 X1 + 21 X2 + 12X3 > 0 …………(8)

• Substituting the values of Xi, we get

• 15 (0.1) + 21 (0.75) + 12 (-1.75) = 0.975%
• Because this is a positive number an arbitrage portfolio has indeed
been identified.
 The Arbitrage Portfolio

• Buy Rs.12,000,000 x (0.1) of Stock 1 = Rs.12,00,000

• Buy Stock 2 = 12,000,000 x (0.075) = Rs.9,00,000
• Total = Rs.21,00,000
• Sell Stock 3 Rs.12,000,000 (-0.175) = Rs.21,00,000
• Additional Fund Required = Rs. Nil
 The Arbitrage Portfolio
Investment Return Risk
Beginning End (beta)

Long position of Stock 1 -Rs12,00,000 Rs13,80,000 Rs 1,80,000 0.090

Long position of Stock 2 -Rs9,00,000 Rs10,89,000 Rs 1,89,000 0.225

Short position of Stock 3 Rs21,00,000 -Rs23,52,000 -Rs 2,52,000 -0.315

Arbitrage portfolio 0 Rs1,17,000 Rs1,17,000 0

 How an arbitrage portfolio affects on investor’s position:

Descriptions Old Portfolio +Arbitrage = New Portfolio

Portfolio

Weight
X1 0.333 0.100 0.433

X2 0.333 0.075 0.408

X3 0.333 -0.175 0.158

Property
rP 16% 0.975% 16.975%

bP 1.9 0.000 1.9

P 11 small Approximately 11.0

 Two Factors Model

• In the case of two factors, denoted I1 and I2 and

representing, for example, industrial production
and inflation, each security will have two
sensitivities i1, and i2.
• Thus security returns are generated by the
following factors model:
• Ri = αi + i1I1 + i2I2 + ei …………. (9)
• And the APT equilibrium
• E(Ri) = Rz + i1λ1 + i2 λ 2 ……….(10)
 Consider a situation in which there are four securities that
have the following expected returns and sensitivities:
Stock E(Ri) βi1 Βi2
1 15% 0.9 2.0
2 21 3 1.5

3 12 1.8 0.7
4 8 2.0 3.2

• In addition, consider an investor who has Rs.5,00,000 invested in each of the

securities. Thus the investor has initial wealth of Rs.20,00,000. Are these
securities priced in equilibrium?
 Arbitrage Portfolios

• An arbitrage portfolio must have weights that satisfy the

following equation:

• X1 + X2 + X3 + X4 = 0 ……………………. (11)
• 0.9X1 + 3X2 + 1.8X3 + 2X4 = 0 ……….….. (12)
• 2X1 + 1.5X2 + 0.7X3 + 3.2X4 = 0 ………… (13)
• Let X1 = 0.1 and then solving the equation, We get X2 = 0.088, X3 = -
0.108 and X4 = -0.08.

• Expected return on this arbitrage portfolio reveals that

• RP = (0.1 x 15%) + (0.088 x 21%) + (-0.108 x 12%) + (-0.08 x 8%) = 1.41%
>0

• Hence, an arbitrage portfolio has been identified.

• Investors will continue to create such arbitrage portfolios until

equilibrium between expected returns and sensitivities is reached as
follows:

• E(Ri ) = Rz + i11 + i22 ………….. (14)

• This is the equation of a plane instead of a line.

 Two Factors Model:
Consider the expected returns and factors sensitivities of
following three assets represent an equilibrium situation
• Find the equation of the plane that must describe equilibrium
return. .
Illustrate the arbitrage opportunities that would exist, if a
portfolio called D exists with the following properties were
observed.

• E(RD)= 15%, D1 = 0.6, D2 = 0.6

• APT Equilibrium Equation:

• Ri = 0 + i11 + i22

• 15 = 0 + 1.01 + 0.62 ….……. (1)

• 14 = 0 + 0.51 + 1.02 ……….. (2)
• 10 = 0 + 0.31 + 0.22 .….…… (3)
• Solving these three equations, we get,
• 0 = 7.75, 1 = 5 and 2 = 3.75.

• Therefore, APT Equilibrium Equation is:

• Ri = 7.75 + 5i1 + 3.75i2

• Compare portfolio D with a portfolio (Call it E) constructed by placing
W1 of the funds in A, W2 in B and W3 in C.
• W1 + W2 + W3 = 1 …………. (1)
• Or W1 = 1 – W2 – W3
• W1A1 + W2B1 + W3C1 = 0.6………. (2)
• W1A2 + W2B2 + W3C2 = 0.6 ………. (3)
• Solving the above equation we get W1 = W2 = W=1/3
• Thus
• E(Ri )= 1/3 (15) + 1/3 (14) + 1/3 (10) = 13
• Alternatively,
• E(Ri )= 7.75 + 5 (0.6) + 3.75 (0.6) = 13

• D1 and D2 are equal to E1 and E2. Hence return should be equal.
But since RD > RE.
• Arbitrage Opportunity is there:

• Buy portfolio D while sell on an equal amount of portfolio E short.

 Riskless Arbitrage

Initial CF End CF Return βi1 βi2

Short Rs100 -113 -13 -0.6 -0.6

Portfolio E
Long -Rs100 +115 15 +0.6 +0.6
Portfolio D
Arbitrage 0 +2 +2 0 0
Portfolio
Review Problems:

Given the three efficiently diversified portfolios in the table with

expected return Ri and sensitivity factors i1 and i2, what is the
equation of the plane in Ri, i1, and i2 space defined by these
portfolios?

Portfolio E(Ri) βi1 βi1

A 14 0.8 0.8

B 10.8 0.6 0.4

C 11.2 0.4 0.6

• The information can be used to derive the
equation for the risk-return plane by
substituting Ri and i

• 14 = 0 + 0.81 + 0.82 …….. (i)

• 10.8 = 0 + 0.61 + 0.42 ……… (ii)
• 11.2 = 0 + 0.41 + 0.62 ……….. (iii)
• Equation (i) - (ii)  3.2 = 0.21 + 0.42 ……… (iv)
• And Equation (iii) - (ii)  0.4 = –0.21 + 0.22 …. (v)
• 0.62 = 3.6 or 2 = 6
• Therefore, 0.21 = 3.2 – 0.4 = 3.2 – 2.4 = 0.8 or 1 = 4

• From Equation (1)

• 14 = 0 + 0.8(4) + 0.8 (6)
• 14 = 0 + 3.2 + 4.8 = 0 + 8
• 0 = 6
• Hence, the required equation E(Ri) = 6 + 4i1 + 6i2
 K. Sundaram owns a portfolio with the
following characteristics:
Stock βi1 βi1 Wi Expected
Reurn
1 1.4 2.5 0.3 13%

2 0.9 1.6 0.3 18

3 1 0.8 .2 10
4 1.3 2 0.2 12
• Assume that returns are generated by a two-
factor model and Sundram decides to create an
a arbitrage portfolio by increasing the holding of
security B by 0.05.

• What must be the weights of the other three

securities in Sundram’s Portfolio?
• What is the expected return on the arbitrage
portfolio?
•
Q.3. Assuming a one-factor model of a firm: Ri = 7% + iF
Consider three well-diversified portfolios (zero non-factor risk). The
expected value of the factor is 10%.

Portfolio βi Ri

A 0.75 14.5%

B 1 15.0%

C 1.5 22.0%
• Is one of the portfolios expected return not in line
with the factor model relationship?
• Can you construct a combination of the other
two portfolios that has the same factor sensitivity
as the out of line portfolio?
• What is the expected return of the combination?
• What action would you expect investors to take
with respect to these three portfolios?
• RA = 7% + (0.75) 10% = 14.5%
• RB = 7% + (1) 10% = 17%
• RC = 7% + (1.5) 10% = 22%
• Thus portfolio B’s expected return is “Out of
Line”.
• Assume XA is the weight in A
• And XC is the weight in C
• Therefore,
• XA (0.75) + XC (1.5) = 1
• XA + XC = 1
• Or XC = (1 – XA)
• Or XA (0.75) + (1 – XA) (1.5) = 1
• Or (0.75) XA + 1.5 – 1.5 XA = 1
• Or XA (1.5 – 0.75) = 1.5 – 1 = 0.5
• XA = 0.5/0.75 = 2/3
• XC = 1/3
 Assume that the following two index model describes returns:
Ri = RZ + i1F1 + i2F2 + ei
Assume that the following three well diversified portfolios are observed:

Portfolio E(Ri) i1 i2

A 12 1 0.5
B 13.4 3 0.2
C 12 3 -0.5
• Find the equation of the plane that must describe equilibrium returns.
• Illustrate the arbitrage opportunities that would exist if a portfolio called D
with the following properties were observed.
• E(RD ) = 10, D1 = 2, D2 = 0
• The equation of the plane
• E(Ri ) = 0 + i11 + i22

• Substituting the values we get three equations:

• 12.0 = 0 + 1 + 0.52
• 13.4 = 0 + 3 1 + 0.22
• 12.0 = 0 + 3 1 – 0.52
• We find, 0 = 10, 1 = 1, 2 = 2

• Therefore, the equation of the plane:

• E(Ri ) = 10 + 1i1 + 2i2
• Let one portfolio E lie on the same plane with E1 = 2 and
E2 = 0 and the portfolio is comprised of three portfolios
with weights W1, W2 and W3. Then

• W1 + W2 + W3 = 1 …………. (1)
• Or W1 = 1 – W2 – W3
• W1A1 + W2B1 + W3C1 = 2………. (2)
• W1A2 + W2B2 + W3C2 = 0 ………. (3)

• Solving we get: W1 = 0.5, W2 = 0, W3 = 0.5

• Therefore, Expected Return of E = 0.5 x 12 + 0 x 13.4 +
0.5 x 12 = 12
• Alternatively,
• RE = 10 + 2 + 0 = 12
• Thus, we have found: E(RD)= 15, D1 = 2, D2 = 0
• E(RE ) = 12, E1 = 2, E2 = 0
 Arbitrage Portfolio:
Initial End CF R βi1 βi2
CF
Long E -100 112 12 2 0
Short D +100 -110 -10 -2 0
Investm 0 +2 0 0
ent
• Assume that security returns are generated by a factor
model in which two factors are pervasive. The
sensitivities of two securities and of the riskfree asset of
each of the two factors are shown below, along with the
expected return on each security:

Security βi1 βi2 E(Ri)

A 0.75 0.6 18%

B 1.75 1.2 24
RF 0 0 12
• If Mr. M has Rs.10,000 to invest and sells short
Rs.5,000 of security B and purchase Rs.15,000 of
security A, find out P1 and P2?

• If Mr. M now borrows Rs.10,000 at riskfree rate and

invest the proceeds of the loan with his original capital
of Rs.10,000 in securities A and B in the same
proportions as described in Part (a), what is the P1
and P2 now? What is E(RP) = ?
• WA = 1.5, WB = -0.5,
• P1 = (1.5) (0.75) + (-0.5) (1.75) = 0.25
• P2 = (1.5) (0.6) + (-0.5) (1.2) = 0.3
• Initial Proportion : 1.5 : (-0.5) i.e., 3 : (-1)

• If Short Sale of B for Rs X and loan for Rs 10,0000, total amount

available for investment in A = (10,000 + 10,000 + X)
• B=-X
• Thus (10,000 + 10,000 + X) = 3X
• Therefore, X = 10,000

• Now Weight: WA = 3, WB = -1, W (risk free asset) = -1

• βP1 = (3) (0.75) – (1) (1.75) – (1) (0) = 2.25 – 1.75 = 0.5
• β P2 = (3) (0.6) – (1) (1.2) – (1) (0) = 1.8 – 1.2 = 0.6
• RP = (3) (18) – (1) (24) – (1) (12) = 54 – 24 – 12 = 18%
•