Beta is an estimate of how a stock co-moves with the broader market. In Saudi markets, beta estimation carries additional complexity due to liquidity differences, free float effects, and periodic market structure changes.
The goal is not to defend a single method but to choose an approach that is appropriate for the valuation use case - and to document that choice clearly.
Beta Inside CAPM
K_e = R_f + (β × ERP)
Beta is a multiplier. If the estimate is unstable, cost of equity inherits that instability, and valuation quality suffers as a result.
Why Regression Beta Can Be Unreliable in Saudi Markets
Regression beta may be noisy when:
- Trading is thin relative to the index - price moves reflect liquidity, not systematic risk
- The company has had structural changes in leverage, business mix, or ownership
- The observation period spans a market regime shift (opening to foreign investors, index reclassification, etc.)
This does not mean regression beta is always wrong. It means you should test its stability before relying on it.
Practical Regression Hygiene
- Test multiple windows - compare 2-year and 5-year estimates
- Compare daily vs. weekly returns for thinly traded names; weekly often reduces noise
- Consider an adjusted (Blume-adjusted) beta when the raw regression is extreme: β_adjusted = (2/3 × β_raw) + (1/3 × 1.0)
- Document the window, frequency, and index used
- Always show a beta sensitivity alongside the selected estimate
Leverage Adjustment and the Zakat Tax Assumption
To remove the effect of capital structure from a peer’s beta, unlever it first:
β_unlevered = β_levered / [1 + (1 − t) × (D/E)]
Then relever to your company’s target capital structure:
β_levered = β_unlevered × [1 + (1 − t) × (D/E)]
Important: For Zakat payers, a conventional interest tax shield typically does not apply. Set t = 0 in both steps unless you have a clearly documented reason to assume otherwise.
Bottom-Up Beta Workflow
Bottom-up beta is often more stable and better suited for decision-grade valuation:
- Select a peer set by business risk (same sector, comparable operating leverage)
- Unlever each peer’s regression beta using their actual D/E and tax rate
- Take the median unlevered beta across the peer set
- Relever using your company’s target capital structure
This approach works even when there is limited trading history, and it reduces the noise from single-stock regression.
Illustrative Example
Assume β_unlevered (peer median) = 0.9, target D/E = 0.5, Zakat payer (t = 0):
β_levered = 0.9 × [1 + (1 − 0) × 0.5] = 0.9 × 1.5 = 1.35
A 0.2 sensitivity around this estimate is reasonable given model uncertainty.
This article is for educational purposes only. It explains valuation concepts and modeling choices. It is not investment advice, a recommendation, or a financial promotion. Examples use illustrative inputs. Verify all data using primary sources and consult an appropriately licensed professional before making financial decisions.
CFA® and Chartered Financial Analyst® are registered trademarks owned by CFA Institute. This content is not endorsed by or affiliated with CFA Institute.
Abdul Gaffar Mohammed, CFA
Treasury & Investment Professional