Skip to content
Corporate & Investment Banking My progress Français

Home / Corporate Finance / Markets and risk / Module 15

15Risk and diversification

What risk is, how it is measured, and why only part of it earns a reward

What risk is, how it is measured, and why only part of it earns a reward.

Reading time
20 min
Sections
14
Exercises
3
Recall questions
6
Quiz questions
12
Simulators
2

What you will be able to do

  • Compute the expected return and volatility of an investment from a table of scenarios.

  • Split a share's risk into a market part and a specific part, and read a beta correctly.

  • Calculate the risk of a two-asset portfolio and explain why it can be below the risk of either asset.

  • Use the efficient frontier, the risk-free asset and the Sharpe ratio to compare portfolios.

  • Explain to a manager why diversifying the company's activities does not, by itself, create value.

1Risk is the spread of outcomes around what you expect

Before measuring anything, decide what is being measured. Finance takes a long list of dangers and reduces them to one question: how far can the result land from the average?

Many dangers, one measure

Many risks reach the investor as one thing

  • Business risk: sales, margins or costs differ from the plan.

  • Financial risk: debt makes the owners' share of the outcome more extreme (CF13).

  • Liquidity and credit risk: cash cannot be raised, or a customer or bank fails to pay.

  • Market risks: exchange rates, interest rates, commodity prices and inflation move.

  • Event risks: cyber attack, fraud, supply chain break, pandemic, climate damage, regulation.

  • Systemic risk: a shock in one part of the financial system spreads to the rest.

Managers see different problems with different owners. An investor sees one: the value of the investment ends up different from what was hoped. A shock reaches value in two ways. It changes the cash flows, or it changes the rate at which they are discounted.

In 2022 inflation rose and central banks raised rates. Many firms saw no change in near-term cash flows, yet their share prices fell, because the discount rate had moved. Shares and bonds fell together in many markets.

Finance counts pleasant surprises as risk too

In finance, risk means dispersion: results above or below expectations. A share that might gain 40% or lose 40% is riskier than one that might gain 8% or lose 2%. People fear the downside most, and the last section of this part looks at it. The basic measure treats both sides alike.

Two numbers summarise a risky bet

The expected return is the average outcome, weighted by how likely each one is

List the outcomes for a year and give each a probability. The expected return is each outcome times its probability, added up. It is not a forecast, but the centre of gravity of what might happen.

E(r) = p₁ × r₁ + p₂ × r₂ + … + pₙ × rₙ

Volatility measures the typical distance from that centre

Take each outcome's distance from the expected return and square it, so that misses above and below do not cancel. Weight the squares by probability and add them: the variance. Its square root is the standard deviation, called volatility, in the same unit as the return.

Variance σ² = p₁ × (r₁ − E(r))² + … + pₙ × (rₙ − E(r))² Volatility σ = √σ²

Volatility misses the shape of the bad outcomes

Real returns have fatter tails than the bell curve suggests

Over long periods, shares have on average paid more than bonds, and bonds more than cash, with years of heavy loss along the way. Returns look like a bell curve, but extreme days occur more often than a bell curve allows. Volatility, built from squared distances, says little about the worst cases. It also moves in clusters: option prices (the VIX in the United States) and models such as GARCH track it as it changes.

Three downside measures look only at what hurts

  • Maximum drawdown is the largest fall from a peak to a later trough. It is what an investor actually lived through.

  • Value at Risk (VaR) gives a loss that should not be exceeded with a stated confidence over a stated period, for example 95% over one month.

  • Expected shortfall is the average loss on the days that are worse than the VaR threshold. It answers the question VaR avoids: how bad is bad?

A loss needs a bigger gain to repair it

If Harmony's share falls from 12.00 to 8.40, it has lost 30%. To return to 12.00 it must gain 42.9%, because the gain starts from a smaller base.

Your cousin Nadia teaches primary school. She says: if a fund loses 30% one year and gains 30% the next, I am back where I started. Explain, using 100 euros, why she is not, and how much it would have to gain.

      Start with 100. A fall of 30% leaves 70. A gain of 30% on 70 adds 21, giving 91. She is 9 down. To return to 100 from 70 the fund needs 30 on a base of 70, which is 42.9%.

      1

      1. Harmony's share returns +62%, +10% or −40% with probabilities 20%, 60% and 20%. Its expected return is…

      2. A fund holds 10 million of Harmony shares. Monthly volatility is 9.24% and expected return 0.867%. Its one-month 95% VaR is about…

      3. Nadia's fund loses 30% one year and gains 30% the next. Starting from 100 she has…

      2Beta measures how much of a share's risk comes from the market

      A share's price moves for two kinds of reason. Some are shared with every other company, and some belong to this company alone. The split between the two is the most useful idea in this module.

      A share's risk has a market part and a specific part

      Some shocks hit everybody and some hit only Harmony

      A fall in consumer confidence, a rise in interest rates or a recession affects almost every company: market risk, also called systematic or non-diversifiable. A recall of a faulty battery batch or the loss of a large dealer affects Harmony alone: specific risk, also called idiosyncratic or diversifiable.

      To separate them, plot Harmony's return against a broad market index over many past periods. The fitted straight line is the market model. Its slope is the beta. The distance of each point from the line is the specific part.

      r(Harmony) = α + β × r(market) + ε

      Here α is the average return the market does not explain and ε is the specific surprise, unrelated to the market's move. Because the two parts are unrelated, their variances add.

      Total variance = β² × σ²(market) + σ²(specific)

      Beta is the slope: how much the share moves when the market moves

      A beta of 1.34 means a 10% market move goes with a 13.4% move in Harmony

      When the market gains 1%, a share with a beta of 1.34 gains 1.34% on average, and falls 1.34% when the market falls 1%. Above 1 amplifies the market, below 1 dampens it, and the market itself has a beta of 1. Statistically, beta is the covariance of the share with the market divided by the market's variance.

      β = Covariance(share, market) ÷ Variance(market) = ρ × σ(share) ÷ σ(market)

      The second form uses the correlation ρ, between −1 and +1, which says how closely two returns move together. For Harmony ρ = 0.67, so β = 0.67 × 32 ÷ 16 = 1.34.

      The beta of a portfolio is the weighted average of its parts' betas: 30% in a share with beta 1.34 and 70% in shares with beta 0.8 gives 0.3 × 1.34 + 0.7 × 0.8 = 0.96.

      2.1 Beta from three statistics

      A listed supplier has a volatility of 24%. The market has a volatility of 16%. The correlation between them is 0.5. Compute the supplier's beta. Then split its variance into a market part and a specific part, and give the fraction of its variance that the market explains.

      β = 0.5 × 24 ÷ 16 = 0.75. Total variance 576. Market part 0.75² × 256 = 144. Specific part 432, a volatility of 20.8%. The market explains 144 ÷ 576 = 25%, which equals ρ².

      Beta comes from the business first, then from the debt, and it must be estimated with care

      Five features push a company's beta up or down

      • Cyclicality: sales that follow the economic cycle, as bikes do, raise beta.

      • Operating leverage: fixed costs make profit swing more than sales (CF10).

      • Financial leverage: debt concentrates risk on shareholders, so equity beta exceeds business beta (CF13).

      • Growth: value in distant cash flows is more sensitive to the discount rate.

      • Visibility: long contracts mean fewer surprises.

      Take the simplest assumption, that debt carries no market risk. Equity beta then equals business beta × (1 + (1 − tax) × net debt ÷ equity value). For Harmony the factor is 1 + 0.75 × 30 ÷ 114 = 1.197, so with no debt its beta would be about 1.12. CF25 turns this into a method.

      A beta is an estimate with a wide error band

      The answer depends on the window (two years or five), the return frequency and the index. A standard error of 0.20 would give Harmony a 95% range from 0.95 to 1.73, wide for a number that feeds a discount rate.

      2

      1. Harmony has a beta of 1.34, a volatility of 32% and the market 16%. How much of its variance is specific?

      2. A single-mine company has a beta of 0.8 and a volatility of 60%. Which statement is right?

      3. With a net debt / equity value of 30 / 114 and 25% tax, Harmony's equity beta of 1.34 implies a business beta of about…

      3Diversification works because assets do not move in step

      Put two assets together and the swings of one partly offset the swings of the other. The effect is free: the return is an average, but the risk is lower than an average.

      With two assets, the portfolio is safer than the average of its parts

      An umbrella seller and an ice-cream seller do better together than apart

      One stall sells umbrellas and earns well when it rains. The other sells ice cream and earns well in the sun. The owner of both has a steadier income. Diversification is the same effect applied to investments.

      The return is an average, the risk has an extra term

      With weight w in asset A and 1 − w in B, expected return is the weighted average. Variance is not: it has a third term that depends on the correlation.

      E(p) = w × E(A) + (1 − w) × E(B)

      σ²(p) = w² × σ²(A) + (1 − w)² × σ²(B) + 2 × w × (1 − w) × ρ × σ(A) × σ(B)

      If ρ = +1, volatility is the weighted average of the two volatilities. Any ρ below +1 gives less. At ρ = −1 the swings cancel and a mix with no risk exists.

      Adding a riskier asset can lower the risk of the portfolio

      For two assets the lowest-volatility weight in A is (σ²B − ρσAσB) ÷ (σ²A + σ²B − 2ρσAσB). With ρ = 0.2 it gives 12.2% in Harmony and 87.8% in the utility, with a volatility of 14.5%, below the utility's 15% alone. Adding a little of the riskier asset lowered risk.

      3.1 The least risky mix

      Asset A has an expected return of 9% and a volatility of 20%. Asset B has an expected return of 5% and a volatility of 10%. They are uncorrelated. Find the weight in A that minimises volatility, then the expected return and the volatility of that portfolio. Is it better than holding B alone?

      With ρ = 0 the weight in A is 100 ÷ 500 = 20%. Expected return = 0.2 × 9 + 0.8 × 5 = 5.8%. Variance = 0.04 × 400 + 0.64 × 100 = 80, volatility 8.9%. It beats B alone: higher return (5.8% against 5%) and lower volatility (8.9% against 10%).

      With many assets, specific risk fades and market risk stays

      In a large portfolio the covariances matter and the individual variances do not

      Take n equal holdings, each with volatility 32% and pairwise correlation 0.3. Portfolio variance is the average variance divided by n, plus the average covariance times (1 − 1/n). As n grows the first term vanishes and the second stays.

      σ²(p) = σ² ÷ n + (1 − 1 ÷ n) × ρ × σ²

      Number of shares Portfolio volatility Risk removed compared with one share
      1 32.0% 0 points
      2 25.8% 6.2 points
      5 21.2% 10.8 points
      10 19.5% 12.5 points
      30 18.2% 13.8 points
      100 17.7% 14.3 points
      Very many 17.5% 14.5 points

      Each share has a volatility of 32% and a pairwise correlation of 0.3.

      Most of the benefit arrives with the first ten shares. The floor of 17.5%, equal to 32% × √0.3, is the risk the shares have in common: market risk. What disappeared is specific risk.

      Your friend Camille keeps all her savings in the shares of her employer. She says she knows the firm well, so it is not risky for her. Explain what she is exposed to twice, and why knowing the firm does not remove the risk.

          Her salary and savings depend on the same company, so trouble could cost her job and savings together. That concentrated, specific risk brings no extra return. Knowing how the firm works does not tell her what will happen to it, and a spread of other shares would remove most of the risk at no cost in expected return.

          a two-asset simulator with sliders for the correlation and the weight in Harmony, plotting portfolio volatility against weight with the minimum-variance point marked. A second tab adds shares one at a time with a slider for average correlation. The learner discovers that the curve bends as correlation falls, that the minimum can sit below both volatilities, and that the many-shares curve flattens at a floor set by the correlation.

          1. 3.3 Two assets are perfectly correlated, with volatilities of 10% and 30%. What is the volatility of a 50/50 mix, and what does this say about diversification?

          2. 3.4 In the table above, why does the 30th share remove so much less risk than the second one did?

          3. 3.5 A fund holds 400 shares, all of them in the same industry. Is it well diversified?

          3

          1. Harmony (32%) and the utility (15%), half each, with a correlation of 0.2. The portfolio's volatility is…

          2. In a portfolio of many shares at 32% volatility and 0.3 correlation, why does the 30th share remove so much less risk than the second?

          3. A fund holds 400 shares, all in the same industry. Is it well diversified?

          4Investors choose how much of one best risky portfolio to hold

          With many assets and a safe alternative, the choice facing an investor becomes simple. The hard part is picking the risky portfolio, and everyone ends up picking the same one.

          The efficient frontier keeps the portfolios no investor should refuse

          A portfolio is dominated when another one offers more return for the same risk

          Take Harmony and the utility at a correlation of 0.2 and vary the weight in Harmony. Each weight is one point on a risk and return diagram.

          Weight in Harmony Expected return Volatility Status
          0% 7.0% 15.0% Dominated
          12.2% 7.4% 14.5% Minimum variance
          25% 7.9% 15.1% Efficient
          50% 8.7% 19.0% Efficient
          75% 9.6% 25.0% Efficient
          100% 10.4% 32.0% Efficient

          The utility alone is dominated: the minimum-variance mix has a lower volatility (14.5% against 15.0%) and a higher return (7.4% against 7.0%). The portfolios from the minimum-variance point upwards form the efficient frontier. Which one to pick depends on how much risk the investor can bear. There is no single best point.

          With many assets the frontier moves outwards. Optimisers also magnify estimation error: a small change in an expected return can swing a holding from nothing to a large amount, so practitioners cap weights or start from the market portfolio.

          A risk-free asset turns the curve into a straight line

          Add a safe asset and one risky portfolio beats all the others

          A risk-free asset has a known return over the investor's horizon. Short-term government bills are the usual proxy, and Harmony's risk-free rate is 3.0%. An investor lends at this rate by holding it and, in theory, borrows at it by taking a negative position.

          Mixing the risk-free asset with a risky portfolio gives a straight line. Of all the lines starting at 3.0%, the steepest one that touches the frontier is best, since it offers the most return for any risk. It touches at the tangency portfolio, which, if investors share the same estimates, is the market portfolio: all risky assets in proportion to market value.

          Capital market line: E(p) = r(f) + [E(market) − r(f)] ÷ σ(market) × σ(p)

          The slope is the Sharpe ratio: (expected return − risk-free rate) ÷ volatility, the standard way to compare portfolios and funds of different risk.

          4.1 Two funds, one ranking

          Fund X returned 9% with volatility 14%. Fund Y returned 8% with volatility 10%. The risk-free rate was 3%. Compute each Sharpe ratio and say which fund rewarded investors better for risk. What else must you check before calling the manager skilled?

          Sharpe of X = 6 ÷ 14 = 0.43. Sharpe of Y = 5 ÷ 10 = 0.50. Y rewarded investors better for the risk, though its return was lower. Check the length of the record, whether a high beta in a rising market explains the return, the fees, and whether luck can be ruled out.

          For a company, the lesson is that only market risk earns a premium

          Harmony's specific risk is risk that investors can remove for nothing

          Harmony's share offers 10.4% for a volatility of 32%. On the capital market line the same 10.4% needs only (10.4 − 3.0) ÷ 0.344 = 21.5%. The extra 10.5 points of volatility come from Harmony's specific risk (23.8% on its own, because variances add, not volatilities), which an investor holding the market portfolio does not carry, so nobody pays a premium for it. The market prices Harmony by its beta: 3.0% + 1.34 × 5.5% = 10.4%.

          CF16 turns this into the security market line, and CF25 builds the cost of capital from it. The principle: a project's discount rate depends on how much of its risk moves with the market.

          A manager cannot create value by diversifying on behalf of shareholders

          Suppose Harmony bought an umbrella maker to smooth its results. Shareholders can already hold both in their own portfolios. If Harmony pays too much, or runs two unrelated businesses worse than specialists would, they lose. Diversifying inside the company pays only if it brings something shareholders cannot copy: cheaper financing, a shared capability, a tax effect, or protection from costly default (CF36).

          Companies do manage some risks, for reasons unrelated to shareholders' appetite: to avoid distress, keep investment funded, protect covenants (CF43).

          Passive funds hold the market at low cost. Active funds try to beat it and must first earn back higher fees. Funds are judged by the Sharpe ratio, the Treynor ratio (excess return per unit of beta) and Jensen's alpha (return above what beta predicts).

          Léa, Harmony's CEO, proposes buying a company in an unrelated sector so that earnings fluctuate less, saying investors will pay more for steadier profits. Explain in plain terms why investors who own other shares may not pay for that, and what she would have to show to justify the purchase.

              Investors who hold many shares already see Harmony's swings partly cancelled elsewhere, so they gain little from steadier profits, and they can buy the other business themselves more cheaply than Harmony can with a premium. Léa would need to show something shareholders cannot copy: real cost or financing savings, a capability worth more combined, or lower risk of costly distress.

              a risk and return map with random portfolios of Harmony and other assets, the frontier on top, and a risk-free slider. The capital market line pivots and touches the frontier at a new tangency point. A second slider for risk appetite moves a marker along the line. The learner discovers that appetite changes the mix and not the risky portfolio, and that Harmony alone sits to the right of the line.

              1. 4.4 What do a very cautious and a very bold investor hold in common?

              2. 4.5 A fund has beaten the market's Sharpe ratio over five years. Give two reasons this may not prove skill.

              3. 4.6 Why must an active fund charging 1.5% a year beat the index before it merely matches it?

              4

              1. The market expects 8.5% with 16% volatility; the risk-free rate is 3.0%. A cautious investor puts 50% in the market. Her portfolio offers…

              2. Fund X returned 9% with 14% volatility, fund Y 8% with 10%, and the risk-free rate was 3%. Which rewarded risk better?

              3. Why does diversifying Harmony into an unrelated business not create value for its shareholders by itself?

              Summary and cheat sheet

              The module’s formulas

              Tap a term to see what it means and its value at Harmony.

              Volatility of a two-asset portfolio

              σ²(p) = + + 2w(1 − w)σ(A)σ(B)

              50/50 mix: 19.0% volatility, against a plain average of 23.5%.

              Splitting a share’s variance

              = +

              Beta

              = × /

              The market prices Harmony by its beta: 3.0% + 1.34 × 5.5% = 10.4%.

              Sharpe ratio

              = ( − ) /

              Idea What to remember
              Risk in finance Dispersion of outcomes around the expected value, up and down.
              Expected return and volatility Probability-weighted average, and the square root of the weighted squared distances. Divide annual volatility by √12 for monthly.
              Downside measures Drawdown, VaR and expected shortfall. A 30% fall needs a 42.9% gain.
              Market versus specific risk Variance = β² × market variance + specific variance. Only the market part cannot be diversified.
              Beta ρ × σ(share) ÷ σ(market). Weighted average for portfolios. Driven by cyclicality, operating and financial leverage.
              Two assets Risk is below the average of the volatilities whenever ρ < 1.
              Many assets Specific risk fades, market risk is a floor.
              Capital market line Risk-free asset plus market portfolio; slope is the Sharpe ratio; appetite sets the mix.
              For the CFO Only market risk earns a premium, so diversifying for shareholders creates no value.

              The one-sentence version

              Risk is the spread of possible outcomes, most of a share's spread can be diversified away at no cost, and so investors are paid only for the part that moves with the market, which beta measures.

              Key terms

              Term Plain-language meaning
              Volatility Typical size of swings in a return, as an annual standard deviation.
              Correlation A number from −1 to +1 saying how closely two returns move together.
              Beta How much a share moves, on average, when the market moves 1%.
              Specific risk Risk peculiar to one company, removable by diversification.
              Efficient frontier Portfolios with the highest return for each level of risk.
              Sharpe ratio Excess return over the risk-free rate, divided by volatility.
              Value at Risk A loss not exceeded at a stated confidence over a stated period.

              Going further

              • Vernimmen, Quiry, Le Fur — Finance d'entreprise (Dalloz), chapters on the risk of a financial security and on risk and portfolio.

              • Brealey, Myers, Allen — Principles of Corporate Finance, chapters on risk, return and portfolio theory.

              • Markowitz (1952) — Portfolio Selection, The Journal of Finance.

              Independent educational material. Not investment, legal or tax advice. All companies and figures in the running case are fictional.