Efficient Portfolios I We will start by studying the characteristics of individual assets and portfolios in term of risk and return focusing on the case of two-assets. Combination of Two risky Assets: Short Sales not Allowed From last lecture, we have seen that the expected return on a portfolio of two assets A and B is given by: 𝑅! = 𝑥! 𝑅! + 𝑥! 𝑅! where 𝑥! = Fraction of the portfolio held in asset A 𝑥! = Fraction of the portfolio held in asset B 𝑅! = expected return on the asset A 𝑅! = expected return on the asset B 𝑅! = expected return on the portfolio P In addition we require the investor to fully invest. That is: 𝑥! + 𝑥! = 1 or 𝑥! = 1 − 𝑥! and substituting 𝑥! = 1 − 𝑥! in the Equation of the expected return, we get: 𝑅! = 𝑥! 𝑅! + (1 − 𝑥! )𝑅! The expected return on the portfolio is a weighted average of the expected returns on the individual securities. As discussed in the previous lecture, the standard deviation of the return on the twoassets portfolio is: 𝜎! = 𝑥!! 𝜎!! + 𝑥!! 𝜎!! + 2𝑥! 𝑥! 𝜎!" !/! where: 𝜎! = standard deviation of the return on the portfolio 𝑃 𝜎!! = variance on the return of the return on asset 𝐴 𝜎!! = variance on the return of the return on asset 𝐵 𝜎!" = covariance between 𝐴 and 𝐵 Substituting 𝑥! = 1 − 𝑥! in the standard deviation and noting that the correlation between A and B, which is defined as: 𝜎!" 𝜌!" = 𝜎! 𝜎! can also be written as: 𝜎!" = 𝜌!" 𝜎! 𝜎! The standard deviation of the portfolio with two assets will become: 𝜎! = 𝑥!! 𝜎!! + 1 − 𝑥! ! 𝜎!! + 2𝑥! (1 − 𝑥! )𝜌!" 𝜎! 𝜎! !/! We consider several cases where the correlation coefficient 𝜌!" is +1, 0, −1 and the intermediate cases (that is when−1 < 𝜌 < 1). In doing so, we consider an example based on two stocks (C and S) that are assumed to have the following characteristics: Asset C Asset S Expected Return (𝑅) 14% 8% Standard Deviation (𝜎) 6% 3% The stock C has a higher expected return and a higher risk than stock S. (Note that the notation of the two assets, in the remaining part of the lecture, will therefore change from A and B to C and S). Perfect Positive Correlation (𝝆 = 𝟏) In this case the variance will be: 𝜎!! = 𝑥!! 𝜎!! + 1 − 𝑥! ! 𝜎!! + 2𝑥! (1 − 𝑥! )𝜎! 𝜎! -------------------------------------------------------------Since this equation has the form 𝑎! + 2𝑎𝑏 + 𝑏 ! = (𝑎 + 𝑏)! -------------------------------------------------------------It follows that: 𝜎!! = 𝑥! 𝜎! + 1 − 𝑥! 𝜎! ! and the standard deviation is: 𝜎! = 𝜎!! = 𝑥! 𝜎! + 1 − 𝑥! 𝜎! (1) While the expected return on the portfolio is: 𝑅! = 𝑥! 𝑅! + (1 − 𝑥! )𝑅! (2) Thus, with the correlation coefficient equal to unity, both risk and return of the portfolio are linear combinations of the risk and return of each security. The form of Equations (1) and (2) means that all combinations of two securities that are perfectly positively correlated will lie on a straight line in risk / return space. To see this, first solve for 𝑥! in Equation (1): 𝜎! = 𝑥! 𝜎! + 𝜎! − 𝑥! 𝜎! or 𝑥! 𝜎! − 𝑥! 𝜎! = 𝜎! − 𝜎! and 𝜎! − 𝜎! 𝑥! = 𝜎! − 𝜎! Substituting this expression in Equation (2) (the expected return) we find that: 𝜎! − 𝜎! 𝜎! − 𝜎! 𝑅! = 𝑅! + 1 − 𝑅! 𝜎! − 𝜎! 𝜎! − 𝜎! or 𝜎! 𝜎! 𝜎! 𝑅! = 𝑅! − 𝑅! + 𝑅! − 𝑅! 𝜎! − 𝜎! 𝜎! − 𝜎! 𝜎! − 𝜎! 𝜎! + 𝑅! 𝜎! − 𝜎! and, finally: 𝑅! − 𝑅! 𝑅! − 𝑅! 𝑅! = 𝑅! − 𝜎! + 𝜎! 𝜎! − 𝜎! 𝜎! − 𝜎! which is the equation of a straight line connecting securities C and S in the expected return / standard deviation space. Since this is a general result we can obtain the straight line for our example. In our example, we have: 𝑅! = 8 𝑅! − 𝑅! 14 − 8 = =2 𝜎! − 𝜎! 6−3 𝜎! = 3 Substituting in the general equation we obtain: 𝑅! = 8 − 2×3 + 2𝜎! or 𝑅! = 2 + 2𝜎! We can see this line graphically. (See Figure 5.1) The two extreme cases are when we invest all in security S or all in security C. These are the points C and S. The other points are portfolios where different proportions are invested in security C and S. • Note that since the risk of the portfolio is the weighted average of the risk on the individual assets, there is no reduction in risk from buying both assets. • This means that nothing has been gained by diversifying rather than purchasing the individual assets. Perfect Negative Correlation (𝝆 = −𝟏) The other extreme case is when the two assets move perfectly together, but in the opposite direction. In this case the variance of the portfolio is: 𝜎!! = 𝑥!! 𝜎!! + 1 − 𝑥! ! 𝜎!! − 2𝑥! (1 − 𝑥! )𝜎! 𝜎! Which can be written as: 𝜎!! = 𝑥! 𝜎! − 1 − 𝑥! 𝜎! ! or 𝜎!! = −𝑥! 𝜎! + 1 − 𝑥! 𝜎! ! -------------------------------------------------------------This is because, in general, an expression like (𝑎! − 2𝑎𝑏 + 𝑏 ! ) can be written as (𝑎 − 𝑏)! or (−𝑎 + 𝑏)! -------------------------------------------------------------- The standard deviation is just the square root of the variance, that is: 𝜎! = 𝑥! 𝜎! − 1 − 𝑥! 𝜎! or 𝜎! = −𝑥! 𝜎! + 1 − 𝑥! 𝜎! We need to notice two things: • Either of the above equations holds only when its right-hand side is positive. • The right-hand side of each question is -1 times the other. It follows that there is a unique solution for the return and risk of any combination of securities C and S. It can be shown that if two securities are perfectly negatively correlated, it is always possible to find a combination of these securities that has zero risk. This combination can be found by setting 𝜎! = 0 in one of the equations for the standard deviation of the portfolio: 0 = −𝑥! 𝜎! + 1 − 𝑥! 𝜎! rearranging: 𝑥! 𝜎! + 𝑥! 𝜎! = 𝜎! and 𝜎! 𝑥! = 𝜎! + 𝜎! Therefore, a portfolio with weights 𝑥! = !! !! !!! and 𝑥! = 1 − 𝑥! will have zero risk. Note also, that since: 𝜎! + 𝜎! > 𝜎! > 0 0 < 𝑥! < 1 This means that the zero-risk portfolio will always involve positive investment in both securities. As for the case of 𝜌 = 1, also when 𝜌 = -1 we will have an equation for the location of all portfolio in expected return / standard deviation space. (See seminar exercise). For our example (i.e. with stocks C and S), we will find the following two equations: 2 𝑅! = 10 + 𝜎! 3 or 2 𝑅! = 10 − 𝜎! 3 (Two equations with same intercept but opposite slopes). Notice that since 𝜌 = -1, we shown that it will exist a combination of these securities (C and S) that provides a portfolio with zero risk. Minimum risk occurs, in our example, when 𝜎! 3 1 𝑥! = = = 𝜎! +𝜎! 3 + 6 3 (See Figure 5.2) We can also have a look at the plot for both cases (that is when 𝜌 = 1 and 𝜌 = -1). (See Figure 5.3) It is important to notice that: • The standard deviation reaches its lowest value for 𝜌 = -1 (zero risk). • The lines SC (when 𝜌 = 1) and SBC (when 𝜌 = -1), represent the limits within which all portfolio of these securities must lie. • This mean that an intermediate correlation (−1 < 𝜌 < 1) should produce a curve such as SOC. Intermediate Risk (−𝟏 < 𝝆 < 𝟏) In the case of an intermediate risk, one point that deserves some attention is the point of minimum standard deviation: that is the portfolio that has minimum risk. This portfolio can be obtained by considering the general formula for the standard deviation or risk: 𝜎! = 𝑥!! 𝜎!! + 1 − 𝑥! ! 𝜎!! + 2𝑥! (1 − 𝑥! )𝜎! 𝜎! 𝜌!" !/! To find the value of 𝑥! that minimize the standard deviation, we take the first derivative with respect to 𝑥! , set the derivative equal to zero and solve for 𝑥! , that is: 𝑑𝜎! =0 𝑑𝑥! If we take the first derivative and solve for 𝑥! , we find the general formula: 𝜎!! − 𝜎! 𝜎! 𝜌!" 𝑥! = ! ! 𝜎! +𝜎! − 2𝜎! 𝜎! 𝜌!" or, equivalently, as: 𝜎!! − 𝜎!" 𝑥! = ! ! 𝜎! +𝜎! − 2𝜎!" (since 𝜎! 𝜎! 𝜌!" = 𝜎!" ) Finally, note that when there in no relationship between returns, that is when for 𝜌!" = 0, the formula becomes: 𝜎!! 𝑥! = ! ! 𝜎! +𝜎! Figure 5.5 shows the graph in the expected return / risk space for all possible values of 𝜌. (See Figure 5.5) Efficient Portfolios II So far we have found that: • The lower (closer to -1) the correlation coefficient between assets, the higher the payoff for diversification. • We have found an expression for finding the minimum variance portfolio when two assets are combined in a portfolio. We now move a bit further in the study of the curve along which all combination must lie in the return / risk space. The Efficient Frontier with No Short Sales Suppose we were able to plot all possible risky assets and combination of risky assets in return / risk space. We are interested in finding a set of portfolios that offered: 1) A Higher Return for the same Risk. 2) A Lower Risk for the same Return. Any investor would only be interested in holding these portfolios (also called efficient portfolios). If we can identify these portfolios, all other portfolios could be ignored. Let’s see if we can identify these portfolios by examining the following Figure 5.8. (See Figure 5.8) Consider Portfolios A and C Portfolio C would be preferred by all investors to portfolio A because it offers less risk for the same level of return. Consider Portfolios A and B Portfolio B would be preferred to portfolio A because it offers a higher return for the same level of risk. Note that: • It is impossible to find portfolios which can dominate portfolio C or portfolio B. • An efficient set of portfolios cannot include interior portfolios like A. Let’s consider only portfolios on the exterior points. Consider point D Portfolio D can be eliminated in favour of Portfolio E (which has a higher return for the same level of risk). This is true for every other portfolio as we move from D to point C on the exterior curve. Portfolio C cannot be eliminated Note that Portfolio C is the global minimum variance portfolio. The global minimum variance portfolio is defined as the portfolio that has the lowest risk of any feasible portfolio. Consider point F Portfolio F is dominated by Portfolio E (which has less risk for the same return). This is true for every other portfolio as we move from F to B. Portfolio B cannot be eliminated Note that Portfolio B is the portfolio (usually a single security) that offers the highest expected return of all portfolios. To summarize: The efficient set consist of the curve of all portfolios that lie between the global minimum variance portfolio and the maximum return portfolio. The set is called efficient frontier. Figure 5.9 shows it graphically. (See Figure 5.9) The Efficient Frontier with Short Sales In the stock market, an investor can often sell a security that she/he does not own. This process is called short-selling. Let’s refresh the concept with an example. Example Suppose security X cost £10 at time T and you think it will be worth £5 at time T+1. You can: • At time T borrow the security from your broker (ignoring transaction cost, dividends, etc…) and sell it for £10: You get £10 but you need to return security X to your broker. • At time T+1, if X falls by £5, you can: go to the market and buy security X for £5. Then return security X that you borrowed making a profit of £5. The main reason for short sales is that the short seller expects the price of the share to decline and would like to profit from the decline. Another reason is to decrease the sensitivity of a portfolio to market movements. Short sales make sense of course when an investor expected return is negative, but it can make sense also when expected returns are positive. This is because the cash flow received from short selling one security with low expected return can be used to buy a security with a higher expected return. When we extend the analysis to efficient frontiers of all securities in the case of short sales, the efficient set still starts with the minimum variance portfolio but it has no finite upper bound. Figure 5.12 shows the efficient set when short sales are allowed (curve MVBC) (See Figure 5.12) Note that infinite expected rate of return exists because one can take a short position on a security with low expected return and use the cash flow to buy a security with higher expected return. Efficient Frontier with Riskless Lending and Borrowing So far we have only considered portfolios or risky assets. We will now introduce a riskless asset in our analysis. Note that: • Lending at a riskless rate can be seen as investing in an asset with a certain outcome. • Borrowing at a riskless rate can be seen as selling such an asset short. The first thing to note is that since the return of the riskless asset is fixed (which we call 𝑅! ), the expected return is: 𝐸[𝑅! ] = 𝑅! (Using the property of the expectation operator) Also, since the return is certain, the variance of the riskless asset is zero: 𝜎!! = 𝐸[𝑅! − 𝐸[𝑅! ]]! = 𝐸[𝑅! − 𝑅! ]! = 0 Assumptions: • Investors can lend and borrow as much as they want. • Investors are interested in placing part of their funds in some portfolio A and either lending or borrowing. • X = fraction of funds invested in portfolio A. • (1 - X) = fraction of funds invested in the riskless asset. The expected return on this combination is: 𝑅! = 1 − 𝑋 𝑅! + 𝑋𝑅! (1) And the standard deviation is: 𝜎! = [(1 − 𝑋)! 𝜎!! + 𝑋 ! 𝜎!! + 2𝑋 1 − 𝑋 𝜎! 𝜎! 𝜌!" ]!/! However, since the value of 𝜎!! is zero, it follows that: 𝜎! = [𝑋 ! 𝜎!! ]!/! = 𝑋𝜎! and rearranging for X we get: 𝜎! 𝑋= 𝜎! Substituting X in the expected return Equation (1), we find: 𝜎! 𝜎! 𝑅! = 1 − 𝑅! + 𝑅! 𝜎! 𝜎! or 𝑅! − 𝑅! 𝑅! = 𝑅! + 𝜎! 𝜎! Equation of a straight line. It follows that the combinations of riskless lending or borrowing with portfolio A lie on a straight line in return / risk space, where: • The intercept is the risk-free asset • The slope is the ratio !! !!! !! This line is shown in Figure 5.13. (See Figure 5.13) Note that: • From the left of A: some funds will be in a riskless asset and some in the risky portfolio A (lending). • On A: all funds will be invested in the risky portfolio A. • From the right of A: some funds will be borrowed. The original funds plus the borrowed funds will be placed in portfolio A. Portfolio A has no special property. However, we can move a bit further and find the best portfolio of risky assets for us to hold. To see this point, consider the intersection of the efficient frontier with risky assets and the efficient frontier when a riskless asset is present. (See Figure 5.14) Consider the lines 𝑅! 𝐵 and 𝑅! 𝐴. Combinations along 𝑅! 𝐵 are superior to combinations along 𝑅! 𝐴, since they offer greater return for the same risk. This is to show that we can find the set of best combinations by extending a straight line up from 𝑅! and swinging it up until it touches point G. (See Figure 5.14) Point G is the tangency point between the efficient frontier and a ray passing through the point 𝑅! . It is important to stress that: Portfolio G is the best portfolio of risky assets for an investor to hold, irrespective of her/his attitude towards risk. In fact: 1) If an investor has a higher degree of risk aversion: he/she would buy both the risk-free asset and portfolio G in some combination (ending up in the 𝑅! G line). 2) If an investor has a lower degree of risk aversion: he/she might want to sell the risk-free asset and use the cash flow, in addition to the original funds, to invest in portfolio G (ending up in the GH line). 3) Other investors would hold risky portfolios with the exact composition of portfolio G. What is important is that: All these investors would hold risky portfolios with the exact combination of portfolio G. Thus for the case of riskless lending and borrowing, we have the solution to the portfolio problem: We can identify the best Portfolio G without having to know anything about the investor attitude to risk. The ability to be able to determine the best (or optimum) portfolio of risky assets has a name: It is called the Separation Theorem. Separation Theorem: Definition Each investor will have a utility-maximizing portfolio that is a combination of the risk-free asset and a portfolio of risky assets determined by the line drawn from the riskfree rate of return tangent to the investor’s efficient set of risky assets.