Multi-Factor Equity Pricing Models – CAIA Level II Study Guide

Multi-Factor Equity Pricing Models – CAIA Level II Study Guide

What Is It?

1. What is this topic?
   Multi-factor equity pricing models explain stock returns using multiple systematic risk factors (e.g., size, value, momentum) beyond just market risk (CAPM).
2. How is it tested, applied, or used?
   Tested via factor regression, portfolio construction, and risk attribution. Used in asset management, hedge funds, and risk modeling to enhance returns or hedge exposures.

Why Does the Exam Ask This?

CAIA tests this to assess your ability to: - Decompose equity returns into explainable risk premia. - Construct and interpret factor-based portfolios. - Evaluate model fit, factor significance, and real-world applicability. - Distinguish between ex-ante (theoretical) and ex-post (realized) factor performance.

What Do I Need to Know First?

1. Capital Asset Pricing Model (CAPM) – Single-factor model (market risk).
2. Regression Analysis – Interpreting coefficients, R², and statistical significance.
3. Factor Investing Basics – Fama-French, Carhart, and macroeconomic factors.
4. Portfolio Construction – Long-short factor portfolios, risk parity.
5. Performance Attribution – Brinson model vs. factor-based decomposition.

Topic Snapshot

Multi-factor models extend CAPM by incorporating additional risk factors (e.g., size, value, momentum) to better explain equity returns. In CAIA Level II, this topic bridges quantitative finance and portfolio management, emphasizing how factors drive returns, how to test them, and how to apply them in real-world investing. Mastery is critical for roles in factor-based strategies, risk management, and hedge fund analysis.

Exam / Job / Audit Weighting

• Frequency: High (appears in 2–3 questions per exam).
• Difficulty Rating: Intermediate.
• Question Type: Calculation (factor regression), conceptual (model interpretation), and application (portfolio construction).

Difficulty Level

Intermediate

Must-Know Rules, Formulas, Standards, or Principles

1. Fama-French 3-Factor Model
   [
   R_i - R_f = \alpha + \beta_{mkt}(R_m - R_f) + \beta_{SMB}SMB + \beta_{HML}HML + \epsilon_i
   ]
2. (SMB) = Small Minus Big (size factor).

3. (HML) = High Minus Low (value factor).

4. Carhart 4-Factor Model (Adds Momentum)
   [
   R_i - R_f = \alpha + \beta_{mkt}(R_m - R_f) + \beta_{SMB}SMB + \beta_{HML}HML + \beta_{UMD}UMD + \epsilon_i
   ]

5. (UMD) = Up Minus Down (momentum factor).

6. Factor Significance & Model Fit

7. T-stat > 2 → Factor is statistically significant.
8. R² → % of return variation explained by the model.
9. Alpha (α) → Unexplained return (skill or luck).

Misconceptions

1. "More factors always improve the model."
2. Overfitting occurs if factors lack economic rationale or are not robust out-of-sample.
3. "Factor premiums are stable over time."
4. Factor returns vary with macroeconomic regimes (e.g., value underperforms in low-growth periods).
5. "All factors are equally investable."
6. Some factors (e.g., momentum) have higher turnover and transaction costs.
7. "Alpha is always skill."
8. Alpha can be misattributed due to omitted factors or survivorship bias.
9. "Factor models replace CAPM."
10. They extend CAPM but do not invalidate it; market risk remains dominant.

Common Mistakes

1. Ignoring factor collinearity – Correlated factors (e.g., value and profitability) distort regression results.
2. Misinterpreting factor signs – Negative (HML) beta ≠ "growth stock"; it means the stock behaves like a growth stock.
3. Overlooking transaction costs – Momentum strategies have high turnover; ignoring costs overstates expected returns.
4. Confusing factor exposure with factor return – A stock can have high (SMB) beta but low (SMB) factor return in a given period.
5. Using raw returns instead of excess returns – Factor models require (R_i - R_f), not (R_i).

The Common Trap

Assuming factor premiums are persistent and universal. - Many candidates treat factors as "always working" (e.g., value stocks always outperform). In reality: - Factor premiums are cyclical (e.g., value underperforms in recessions). - Factor definitions vary (e.g., book-to-market vs. earnings-to-price for value). - Factors can disappear or reverse (e.g., momentum crashes).

Terms to Remember

1. Factor Mimicking Portfolio – A long-short portfolio designed to isolate a single factor (e.g., SMB = small-cap long, large-cap short).
2. Factor Loading (Beta) – A stock’s sensitivity to a factor (e.g., (\beta_{HML} = 0.8) means 80% exposure to value).
3. Idiosyncratic Risk – Return variation unexplained by factors (residual in regression).
4. Factor Timing – Dynamically adjusting factor exposures based on macroeconomic conditions.
5. Factor Crowding – When too many investors chase the same factor, reducing its future premium.

Step-by-Step Process

1. Select a Factor Model

• Choose based on purpose:
  • Fama-French 3-Factor → Basic equity risk decomposition.
  • Carhart 4-Factor → Momentum inclusion.
  • Macro-Factor Models → Inflation, GDP growth.

2. Gather Data

• Stock returns (excess of risk-free rate).
• Factor returns (e.g., SMB, HML from Kenneth French’s data library).
• Risk-free rate (T-bills or SOFR).

3. Run Regression

• Use OLS regression to estimate factor betas and alpha.
• Check:
  • T-stats (significance of betas).
  • R² (model explanatory power).
  • Alpha (intercept; should be close to zero for diversified portfolios).

4. Interpret Results

• Positive beta → Stock benefits when factor outperforms.
• Negative beta → Stock benefits when factor underperforms.
• High R² → Factors explain most return variation.
• Significant alpha → Manager skill or model misspecification.

5. Apply to Portfolio Construction

• Factor Tilting – Overweight stocks with desired factor exposures.
• Factor Neutralizing – Hedge unwanted factor risks.
• Long-Short Factor Portfolios – Isolate pure factor returns.

6. Evaluate Model Fit

• Out-of-sample testing – Does the model work on unseen data?
• Economic rationale – Are factors backed by theory (e.g., value = risk premium for distress)?
• Transaction costs – Do factor strategies remain profitable after costs?

Exam Answer Builder

1-Mark Question (Single-Best-Answer MCQ)

What it tests: Recognition of factor definitions. Example Question: Which of the following best describes the "SMB" factor in the Fama-French model? A) Small-cap stocks minus large-cap stocks B) High book-to-market stocks minus low book-to-market stocks C) High-momentum stocks minus low-momentum stocks D) High-beta stocks minus low-beta stocks

Correct Answer: A Key Tip: Memorize factor definitions (SMB = size, HML = value, UMD = momentum).

2-Mark Question (Short Calculation)

What it tests: Factor regression interpretation. Example Question: A stock has the following Fama-French 3-factor regression results: - (\alpha = 0.5\%) - (\beta_{mkt} = 1.2) - (\beta_{SMB} = 0.3) - (\beta_{HML} = -0.7) - (R² = 0.85)

If the market return is 8%, SMB is 2%, and HML is -1%, what is the stock’s expected return (assuming risk-free rate = 2%)?

Solution: [ R_i = R_f + \beta_{mkt}(R_m - R_f) + \beta_{SMB}SMB + \beta_{HML}HML + \alpha ] [ R_i = 2\% + 1.2(8\% - 2\%) + 0.3(2\%) + (-0.7)(-1\%) + 0.5\% = 10.6\% ]

Key Tip: Plug in the numbers carefully; don’t forget alpha!

5-Mark Question (Long Answer)

What it tests: Factor model application and critique. Example Question: A hedge fund manager claims their portfolio has a consistent 2% annual alpha based on the Fama-French 3-factor model. Critically evaluate this claim, discussing potential issues with the model and alternative explanations for the alpha.

Key Points to Include: 1. Model Misspecification – Omitted factors (e.g., momentum, profitability) could explain alpha. 2. Data Mining – Alpha may result from overfitting to historical data. 3. Survivorship Bias – If the portfolio excludes failed stocks, alpha is overstated. 4. Transaction Costs – High turnover (e.g., momentum) may erode alpha. 5. Factor Timing – Alpha could reflect dynamic factor exposure, not skill.

Key Tip: Structure your answer: (1) Define alpha, (2) List issues, (3) Suggest improvements (e.g., Carhart 4-factor, out-of-sample testing).

Case Study Question (Application)

What it tests: Factor-based portfolio construction. Example Question: You are constructing a factor-based equity portfolio. Your benchmark is the S&P 500. How would you design a portfolio to have: - 1.5x exposure to the value factor (HML). - 0.5x exposure to the size factor (SMB). - Neutral exposure to the market factor.

Describe your approach, including stock selection, weighting, and risk management.

Key Points to Include: 1. Stock Selection – Rank stocks by book-to-market (value) and market cap (size). 2. Long-Short Construction –
- Value: Long high book-to-market, short low book-to-market (1.5x leverage).
- Size: Long small-cap, short large-cap (0.5x leverage).
- Market Neutral: Offset market beta with futures or ETFs. 3. Risk Management –
- Monitor factor correlations (e.g., value and size often move together).
- Rebalance quarterly to maintain target exposures.
- Stress-test for factor reversals (e.g., value underperforming for years).

Key Tip: Emphasize practical considerations (costs, liquidity, rebalancing).

This vs That: Multi-Factor Models vs. CAPM

Time-Saver Hack

Quick Factor Exposure Check: - Value Stock? → High book-to-market, low P/E, high dividend yield. - Small-Cap Stock? → Market cap < 25th percentile of universe. - Momentum Stock? → Strong past 6–12 month returns (excluding most recent month). - Negative HML Beta? → Growth stock (low book-to-market). - Negative SMB Beta? → Large-cap stock.

Mini Scenarios

1. Basic Scenario

You run a Fama-French regression on a stock and find: - (\beta_{HML} = 1.2) - (\beta_{SMB} = -0.5)

What does this tell you? - The stock behaves like a value stock (high HML beta) and a large-cap stock (negative SMB beta). - Action: If you want to hedge value exposure, short value stocks or go long growth stocks.

2. Applied Scenario

A portfolio manager claims their fund has a 3% annual alpha using the Carhart 4-factor model. Upon closer inspection, you notice the fund has a (\beta_{UMD} = 0.8) and the momentum factor returned 5% last year.

What’s happening? - The "alpha" may be misattributed momentum exposure. - Action: Re-run the regression with a longer time horizon or test for factor timing.

3. Tricky Scenario

A stock has the following factor exposures: - (\beta_{mkt} = 1.1) - (\beta_{SMB} = 0.4) - (\beta_{HML} = -0.3) - (\beta_{UMD} = 0.2)

The market drops 10%, SMB drops 5%, HML rises 3%, and UMD drops 2%. What’s the stock’s expected return (ignoring alpha)?

Solution: [ R_i = 1.1(-10\%) + 0.4(-5\%) + (-0.3)(3\%) + 0.2(-2\%) = -12.3\% ] What to notice: The stock underperforms the market due to negative HML beta (growth exposure) and positive SMB beta (small-cap exposure) in a down market.

Diagnostic MCQ Bank

Easy (3 Questions)

Question 1

Which factor is NOT included in the Fama-French 3-factor model? A) Market B) Size C) Momentum D) Value

Correct Answer: C Explanation: The Fama-French 3-factor model includes market, size (SMB), and value (HML). Momentum is added in the Carhart 4-factor model. Trap Option: B (size is included, but momentum is not).

Question 2

A stock has a (\beta_{HML} = -0.8). What does this imply? A) The stock is a value stock. B) The stock is a growth stock. C) The stock has high momentum. D) The stock is small-cap.

Correct Answer: B Explanation: Negative HML beta means the stock behaves like a growth stock (low book-to-market). Trap Option: A (positive HML beta = value stock).

Question 3

In a factor regression, what does a high R² indicate? A) The model explains most of the stock’s return variation. B) The stock has high alpha. C) The factors are statistically insignificant. D) The stock is mispriced.

Correct Answer: A Explanation: R² measures the % of return variation explained by the model. Trap Option: B (alpha is unrelated to R²).

Medium (4 Questions)

Question 4

A portfolio has the following factor exposures: - (\beta_{mkt} = 0.9) - (\beta_{SMB} = 0.5) - (\beta_{HML} = -0.2)

If the market returns 5%, SMB returns 3%, and HML returns -1%, what is the portfolio’s expected return (ignoring alpha and risk-free rate)?

A) 4.2% B) 4.7% C) 5.2% D) 5.7%

Correct Answer: B Calculation: [ 0.9(5\%) + 0.5(3\%) + (-0.2)(-1\%) = 4.5\% + 1.5\% + 0.2\% = 6.2\% \quad \text{(Wait, this is wrong!)} ] Correction: [ 0.9(5\%) + 0.5(3\%) +