Choosing Return Horizons and Historical Depth for CAPM Beta
Summary
The document considers whether a stock’s CAPM beta should vary with the investment horizon used to make a capital budgeting decision. The response treats beta estimation as an unstable-parameter problem and suggests matching the return horizon to the project horizon as a straightforward approach. It also says beta can be estimated from daily, weekly, or monthly observations, while emphasizing that the historical estimation window may matter more than sampling frequency. For long horizons, it suggests that a beta near the market level could be a reasonable consideration.
A second response offers an adjusted beta that blends an estimated beta toward one, and notes that company events can shift beta as market conditions change. These are practitioner views, not a settled economic explanation or an empirical comparison. The discussion cites no supporting study for the stated data-frequency convention, and it leaves open how the appropriate lookback period or adjustment should be selected for a particular firm or project.
Key ideas
- Project horizon can motivate estimating beta with returns measured over a matching horizon.
- The discussion presents beta as subject to parameter instability and changing variance and correlation.
- The respondents differ on whether observation frequency matters, while one emphasizes the historical lookback depth.
- An adjusted beta that moves an estimate toward one is offered as a practical heuristic, not a proven rule.
- Company-specific events can alter a stock’s relationship with market movements.
Tags
Full text
# CAPM betas at different horizons # CAPM betas at different horizons When taking capital budgeting decisions appropriate cost of capital should depend on the horizon of the investment. So the beta of a stock, i.e. it's covariance with the market should depend on the horizon of the investment. For some stocks, beta might be declining with maturity. For others, beta might be increasing with maturity. Is there any literature out there that talks about this issues? Why economically and fundamentally, beta for a company at different horizons might differ? ## Answer by markowitz (score 1) https://quant.stackexchange.com/a/26199 If your issue is the holding period "sensibility", I don't have persuasive economic/fundamental motivation about it. It's an interesting question. Anyway in econometric point of view the issue is part of instability parameters problem. If the time horizon for your investment project is, for example, one year, then you need return and beta one year based. Therefore, in simplest way, you have to take yearly data return to estimate the beta, with certain historical depth. However the beta are not (explicitly) time horizon dependent and you can take the monthly/weekly/daily data. In my experience the time frequency are not so important in beta estimate (also if the number of observation change and this tend to suggest, at least in my opinion, weekly or daily data ... but monthly are largely used). In opposite is so important the historical depth. This problem is strictly related with time varying variance and correlation. However the "beta structure" and experience suggest that it move around one and this fact alleviate the problem. You can see for example Blume's technique. Maybe if your time horizon is too long the better choice for beta becomes 1. In any case, for example and at least in the past, Merrill Lynch suggest to use monthly data with 60 obs. In the past I looked for but I'm not found empirical research for support them choice. Hope that helps ## Answer by clocker (score -1) https://quant.stackexchange.com/a/24824 Hmm, I guess you're trying to determine the cost of capital using CAPM. Generally speaking, beta should stabilize for mature companies. Personally, I like to use an adjusted beta, eg: 1/3*current_beta + 2/3 to reflect this. Beta of a stock is just a measure of how it's covariance is affected compared to the market's volatility as you pointed out. Here's a link from Investopedia about calculating beta: http://www.investopedia.com/ask/answers/070615/what-formula-calculating-beta.asp Since it is clearly a function of market dynamics, beta is often subject to the whims of investors, and sometimes rightfully. Mature companies can go either way -- imagine a major lawsuit that affects a well established public company. A high profile case could cause the volatility of the stock to be extremely volatile for many months in a "serene" market causing its beta to jump. Hope that helps.
Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.