Macroeconomic, fundamental and statistical factor models

Macroeconomic, fundamental and statistical factor models

Published September 28, 2026

Macroeconomic, fundamental and statistical factor models

On this page

Learning Outcome Statement

Describe and compare macroeconomic factor models, fundamental factor models, and statistical factor models.

On the exam: The vignette gives a client objective, constraints, benchmark, investment universe, and factor definition or estimation method. The question asks you to describe or compare model types and choose the model with the most implementable portfolio consequence.

A risk model is a map from what moves markets to what your portfolio actually owns. You need to distinguish how each model creates factors, then judge whether its exposures can guide a constrained trade. Historical fit matters, but a portfolio manager also needs factors that can be named, monitored, and acted on.

ModelFactor definitionPortfolio consequence
MacroeconomicEconomic variables or economic surprisesScenario-based exposure controls
FundamentalObservable security or issuer characteristicsTrait-based screens and limits
StatisticalHistorical return or covariance structureData-driven co-movement map

The factor source determines what the investment committee can explain. A model can be statistically impressive and still be difficult to translate into a holding-level decision. Read the first table as a map from factor origin to a portfolio use: economic inputs support scenario controls, issuer traits support screens, and return history describes co-movement. In a vignette, use the stated evidence to identify the factor source first. Then ask whether the resulting exposure can be monitored and changed within the mandate. That second step prevents a label from becoming an unsupported recommendation.

Evidence in the vignetteModel to testPortfolio question
Economic variables or surprisesMacroeconomicCan you set a scenario-based exposure control?
Observable security or issuer characteristicsFundamentalCan you use a trait-based screen or limit?
Historical return or covariance structureStatisticalCan the co-movement map lead to an eligible trade?

A factor model tells you how a security's return is assembled from common forces and a security-specific remainder. Call that return $R_i$, where $i$ identifies the security. The baseline component is $\alpha_i$. For factor $j$, $\beta_{ij}$ measures security $i$'s exposure to the factor realization $F_j$. The leftover movement is $\epsilon_i$.

The model writes that story as:

$$ R_i = \alpha_i + \sum_j \beta_{ij}F_j + \epsilon_i $$

$\alpha_i$ shifts the modeled return as the security's baseline. Each term $\beta_{ij}F_j$ adds a common-factor contribution. A positive exposure moves with a positive factor realization, while a negative exposure moves against it. $\epsilon_i$ holds movement that the chosen factors do not explain.

Use model selection as a portfolio decision, not a contest for historical fit. Balance explanatory power, interpretability, stability, data requirements, and implementability. A model can fit the past well and still fail if its inputs arrive late, its exposures change unpredictably, or no eligible holding can reduce the risk.

Worked example

You are handed a pension fund mandate for a long-only equity portfolio. The committee wants to control unintended style and issuer exposures while staying inside liquid benchmark constituents. The exam asks which model you recommend and what portfolio action it supports. Read the diagram above as a decision sequence: identify the mandate, test whether economic variables answer its risk question, then check for observable traits if they do not. The final branch is not a free choice; it follows from the available factor definition. Whatever model you select, return to the long-only constraint and benchmark universe to confirm that the exposure can be changed with eligible holdings.

Case factGiven
Client typePension fund
ObjectiveControl style and issuer exposures
ConstraintLong-only, liquid holdings
BenchmarkBroad equity index
UniverseBenchmark constituents
Factor informationEconomic variables and characteristics available

The case facts separate the target from the available tools. Style and issuer control points toward observable characteristics, while economic variables remain available as a possible competing factor source. The broad index supplies the comparison point, and the liquid, long-only universe limits the trades that can express a recommendation. Use these details in order: identify the objective, match it to a factor definition, then verify that the proposed adjustment stays inside the mandate. Use this decision path.

Loading diagram…
  1. Start with the objective before naming an algorithm. The objective is control of style and issuer exposures, so the risk question is characteristic-based. This produces the target: control of those characteristic exposures.

  2. Match the factor definition to that target. Observable security and issuer characteristics point to a fundamental factor model because the committee needs direct control language. This produces the model choice: a fundamental factor model.

  3. Map the choice to the constraint and universe. Long-only holdings limited to benchmark constituents let the manager screen, size, or replace eligible securities to change those exposures. This produces an implementable route that stays within the mandate.

  4. Carry the benchmark into monitoring. Compare the portfolio's characteristic exposures with the broad equity index because the committee needs to identify unintended differences. This produces a benchmark-aware monitoring comparison.

  5. Recommend the fundamental factor model. It matches the objective, uses available data, and supports a concrete action through eligible holdings. This produces the final recommendation: a fundamental factor model.

The decision is a fundamental factor model, with characteristic exposures monitored and corrected through eligible holdings.

Changed condition

Keep the fund, universe, constraint, benchmark, and available information unchanged. Change only the objective from controlling style and issuer exposures to controlling inflation and interest-rate surprises.

Case featureOriginalChanged
Client objectiveControl style and issuer exposuresControl inflation and interest-rate surprises
ConstraintLong-only, liquid holdingsLong-only, liquid holdings
BenchmarkBroad equity indexBroad equity index

Only the objective moved. The preferred model becomes a macroeconomic factor model because its factors are defined by economic variables or economic surprises, which match the new risk question. The unchanged constraint and benchmark still require exposures the manager can monitor and express through eligible holdings.

Before reading on, choose the factor definition and portfolio action you would use in the changed case.

Common error

You choose the statistical model solely because it explains historical covariance best. The choice feels objective because the data produced the fit, and the analyst can point to a measured result. The giveaway is that you never connect the model to an economic or security characteristic, benchmark exposure, client constraint, or portfolio action that a portfolio manager could implement.

Check yourself

Practice

A pension fund committee reviews competing risk models for a constrained benchmark portfolio. The models use economic surprises, observable issuer characteristics, or historical covariance. The committee must choose a model that supports an implementable control.

Question 1

A research team defines its factors as inflation surprises, changes in policy rates, and shifts in credit spreads. Which model is being used?

Question 2

A manager defines exposures from issuer leverage, market size, and valuation characteristics that are observable before trades. Which model fits?

Question 3

A risk team extracts factors from the historical return covariance structure and gives them no economic labels. Which model fits?

Question 4

A constrained portfolio has a benchmark and needs a risk control the manager can implement through eligible holdings. Which selection rule is best?

A statistical model explains the portfolio's historical covariance best, but its factors cannot be named or linked to a tradable exposure. What is the best conclusion?

Revision snapshot

Macroeconomic models define factors using economic variables or economic surprises. Fundamental models define exposures using observable security or issuer characteristics. Statistical models extract factors from the historical return or covariance structure.

$$ R_i = \alpha_i + \sum_j \beta_{ij}F_j + \epsilon_i $$

Choose a model by balancing explanatory power, interpretability, stability, data requirements, and implementability. A model that fits historical returns best is not automatically the best model for a constrained portfolio. On the exam, identify the objective, constraint, benchmark, client type, universe, and factor definition, then select the model that supports an implementable portfolio action.

Unlock the full lesson and continue reading

Signup to continue reading this lesson and unlimited access questions, flashcards, AI notes, and more

Scan to download app

Scan to download
UNLOCK FREE ACCESS

Practice Questions: Macroeconomic, fundamental and statistical factor models

  1. 1. A portfolio invests 70% in an asset with 15% volatility and 30% in an asset with 25% volatility. The portfolio's volatility is 15.66%. The correlation between the two assets is closest to:

    • A. 0.50
    • B. 0.25
    • C. 0.86
    Show answer

    Answer: A. 0.50

    0.50 - Portfolio variance is 0.1566² = 0.02452356, while the assets’ individual variance contributions total 0.70²(0.15²) + 0.30²(0.25²) = 0.01665. - The covariance coefficient is 2(0.70)(0.30)(0.15)(0.25) = 0.01575, so correlation is (0.02452356 − 0.01665)/0.01575 ≈ 0.50. 0.25 - Double-counting covariance changes its coefficient from 0.01575 to 0.0315, giving (0.02452356 − 0.01665)/0.0315 ≈ 0.25. - The portfolio variance formula already includes the factor of two for covariance, so doubling that term again understates correlation. 0.86 - Omitting the second asset’s variance contribution of 0.30²(0.25²) = 0.005625 gives (0.02452356 − 0.011025)/0.01575 ≈ 0.86. - Both assets’ individual variance contributions must be subtracted before the remaining variance is attributed to covariance.

  2. 2. Meridian Asset Management is considering launching an actively managed equity fund benchmarked to a broad domestic equity index. The proposed strategy combines value and momentum signals and is expected to hold 80–120 securities. The investment committee requires the manager to distinguish between total portfolio risk and active risk and to evaluate performance after realistic implementation costs. The portfolio manager, Elena Park, presents the forecasts in Exhibit 1. She states that the expected information ratio should be calculated using the portfolio's total expected return in the numerator because clients ultimately receive total return. A risk analyst disagrees and notes that the mandate evaluates skill relative to the stated benchmark. The analyst also warns that the portfolio's beta of 1.08 may generate benchmark-related returns that should not automatically be interpreted as security-selection skill. The risk team models the portfolio using market, value, and momentum factors. Exhibit 2 reports each source's contribution to portfolio variance. The reported factor contributions already incorporate factor exposures, factor variances, and relevant factor covariances. Consequently, the listed contributions can be added directly. Specific risk reflects security-level residual risk after accounting for the common factors. Park's research team backtested the strategy over 15 years. To simplify data collection, it downloaded the companies currently included in the benchmark and applied the strategy retrospectively to those companies throughout the full historical period. The team did not add companies that had failed, merged, or previously been removed from the index. Signal calculations otherwise used only accounting and market data publicly available at each simulated rebalancing date. The researchers selected the value and momentum specification before running the reported backtest and did not search repeatedly across alternative model specifications. Exhibit 3 presents gross simulated active return and estimated annual implementation costs. Commission and bid–ask estimates are based on expected trade size. Market-impact estimates reflect the strategy's forecast assets under management and average daily trading volume. The management fee is charged as a percentage of assets annually. Park argues that only commissions should be deducted because market impact is not shown explicitly on trade confirmations and management fees are not trading costs. The committee instead requires an estimate of client performance net of all economically relevant implementation costs and fees. All returns and risks in the exhibits are annualized using consistent conventions.

    • 3. Northbridge Pension Plan is reviewing instruments and risk measures for a tactical equity allocation. The investment committee is considering an exchange-traded fund that holds a transparent basket of domestic equities. Large financial institutions acting as authorized participants can exchange the published basket for creation units and can redeem creation units for the basket. The ETF's creation basket corresponds to a standard creation unit size, as shown in Exhibit 1. During the review, the ETF's market price rises above its contemporaneous per-share net asset value, as shown in Exhibit 1. There are no taxes, trading restrictions, or material transaction costs for the authorized participant. The committee also asks the risk team to estimate the one-day loss threshold for the plan's separate equity portfolio. Daily returns are assumed to be normally distributed and serially independent, with the mean and standard deviation reported in Exhibit 1. The plan defines value at risk as a positive loss amount that should be exceeded with the stated tail probability. The risk team is instructed to use the variance-covariance method and the supplied one-tailed critical value. The risk team is also asked to estimate expected shortfall using the supplied one-tailed expected shortfall multiplier, to compare the average tail loss with the value at risk threshold. Northbridge allocates risk using component contributions to portfolio volatility. For each asset, component contribution is defined as portfolio weight multiplied by marginal contribution to volatility. Percentage contribution to total volatility is the asset's component contribution divided by the sum of all component contributions. Exhibit 2 reports the required inputs. The committee is concerned that simply equating capital weights with risk contributions could understate the dominance of the higher-volatility equity allocation. A committee member additionally proposes replacing part of the tactical position with a leveraged ETF. The fund seeks to deliver two times the index's return for each trading day before fees and expenses. It resets its exposure at the end of every trading day. The member claims that an investor holding the fund for one month should therefore earn exactly two times the index's one-month return, regardless of the sequence of daily index returns. The risk officer notes that compounding after daily exposure resets can cause the cumulative result to depart from a constant multiple of the index's holding-period return, particularly in volatile markets. All ETF transactions are evaluated using contemporaneous prices. The questions concerning ETF arbitrage, value at risk, expected shortfall, risk contributions, and leveraged returns are independent.

      • 4. Despite credible new information reducing a company's value, an investor makes only a minor revision to her original USD 80 estimate. Her behavior most likely reflects:

        • A. Anchoring and adjustment
        • B. Confirmation bias
        • C. Loss-aversion bias
        Show answer

        Answer: A. Anchoring and adjustment

        Anchoring and adjustment - The original USD 80 estimate serves as an anchor despite credible new information that reduces the company’s value. - Revising the estimate only slightly shows insufficient adjustment away from that anchor. Confirmation bias - Confirmation bias would involve favoring information that supports the USD 80 estimate or discounting contrary evidence. - The question describes a small revision, not selective treatment of the new information. Loss-aversion bias - Loss aversion involves weighing losses more heavily than gains when making decisions. - The question describes how the investor updates a valuation estimate, not a decision driven by a potential loss.

      • 5. A portfolio invests 60% in an asset with 12% volatility and 40% in an asset with 10% volatility. Their correlation is 0.25. The calculated value of portfolio volatility is closest to:

        • A. 8.0%
        • B. 9.2%
        • C. 10.0%
        Show answer

        Answer: B. 9.2%

        9.2% - Portfolio variance is \((0.60^2)(0.12^2)+(0.40^2)(0.10^2)+2(0.60)(0.40)(0.12)(0.10)(0.25)=\mathbf{0.008224}\). - Its square root is 9.07%, making 9.2% the closest choice. 8.0% - Omitting the covariance term gives variance of \(0.005184+0.001600=0.006784\), or volatility of 8.24%, closest to 8.0%. - That omission incorrectly treats the assets as uncorrelated despite their stated correlation of 0.25. 10.0% - Using the second asset’s 10% volatility directly produces this number. - Portfolio volatility must reflect both assets’ weights and their correlation, not just the volatility of the asset with a 40% weight.

      • 6. An investor becomes more attached to inherited shares after taking ownership. Which action is most consistent with this behaviour?

        • A. Maintain the holding because selling would change the status quo
        • B. Assign the holding a value above its observable market price
        • C. Retain the holding because losses would cause disproportionate regret
        Show answer

        Answer: B. Assign the holding a value above its observable market price

        Assign the holding a value above its observable market price - Endowment bias leads the investor to value the inherited shares more highly simply because they now own them. - The valuation above the observable market price reflects attachment to ownership, not new information about the shares. Maintain the holding because selling would change the status quo - Status quo bias favors keeping the inherited shares because doing so avoids a change. - This choice explains inaction, but not why ownership would raise the investor’s valuation of the shares. Retain the holding because losses would cause disproportionate regret - Loss aversion makes a potential loss feel more consequential than a comparable gain. - This choice explains reluctance to sell at a loss, rather than valuing the inherited shares more highly because they are owned.

      Flashcards: Macroeconomic, fundamental and statistical factor models

      1. 1. What are the main features of macro factor model?

        Show answer

        - price risk factors - "not priced" means investors cannot expect to be rewarded for being exposed to that type of risk - systematic risk factors -> will affect even well-diversified portfolios - cannot be avoided - factor sensitivites - can be estimated by regressing historical asset returns on the corresponding historical macro factors

      2. 2. What do fundamental factor models use?

        Show answer

        firm-specific valuation metrics such as PE with standardized sensitivites

      3. 3. What are uses of multifactor models?

        Show answer

        - passive management - active management - rules-based or algorithmic active management

      4. 4. What are the potential benefits of using more risk factors?

        Show answer

        - can isolate risks that have a competitive advantage - select more efficient portfolios

      5. 5. What are the four factors of the Carhart Model?

        Show answer

        market risk, size, value and momentum - extension of the Fama French model

      6. 6. What two risk factors does ICAPM price?

        Show answer

        1. global market index: company's relationship with the local economy relative to the global economy 2. a foreign currency-denominated, wealth-weighted market index: sensitivity to the company's cash flows to changes in its currency value

      7. 7. What are the assumptions of the Arbitrage pricing theory?

        Show answer

        - unsymstematic (firm-specific) risk can be diversified away in a portfolio - systematic risk is priced - returns generated using a factor model - no arbitrage opportunities exist - asset returns are described by a k factor mode

      8. 8. What is arbitrage pricing theory?

        Show answer

        alternative to CAPM, linear model with multiple risk factors priced by the market - weakness: doesn't identify specific risk factors - does not require that one of the risk factors is the market portfolio

      Everything you need for CFA prep

      Get full Oncourse access with lessons, practice questions, flashcards and AI study tools.

      GET STARTED FOR FREE
      75% OFFLimited time offer
      GET 75% OFF