I can’t figure out how the expected market return is calculated. I understand it’s function in CAPM and in determining beta. But I don’t see how any value for expected market return, market covariance with your stock, or variance of the market, that you’re given is reliable. Has someone ever calculated the expected return for each asset available, the SD for each asset, or correlation between every pair of assets? To me this seems like the #1 flaw of CAPM because subjective probabilities are needed to even generate variance!!!
look up grinold-kroner model. it’s all guess work. if everyone could just plug & chug the market would be perfectly efficient.
If you use CAPM, you do not need to compute the correlations for every pair of assets in the market. They are derivable from the betas, so you just need to calculate betas and SDs for every asset in the market. As for the market return, you can try to use a long term average for market returns. We don’t know what the market will return this year, but we know that, on average, the S&P returns about 4 or 5% more than Treasuries. Generally, you start there, and then you can refine that model further by using something like Grinnold/Kahn’s model.
But what if there’s no dividend? I thought the Grinold Kroner model requires a dividend paying stock.
"If you use CAPM, you do not need to compute the correlations for every pair of assets in the market. They are derivable from the betas, so you just need to calculate betas and SDs for every asset in the market. " To calculate the beta’s you’d need the covariance of the stock and the market, and variance of the market. Which is what I’m getting at. The variance would use subjective probabilities. What would be the most reliable non-dividend paying equity valuation tool? Fama-French?
The variance of the market tends to be (substantially) more stable than the expected returns. So the historical variance is not so bad. Covariance is less stable, but still better than the mean. So, yeah, it’s not perfect, but it’s not the same level of subjectivity as guessing the expected returns.
Just use 8%.