High Beta Stocks: Plenty of Risk, Little Reward

Many investors try to add leverage to their portfolios through high beta stocks. The thinking is if the market rebounds, they will get a greater pop with the higher risk/higher return nature of high beta stocks, because we always read that high beta stocks are defined by the multiple by which their price movements have behaved relative to the market. For example, a stock with a beta of 2 "means" that if the market goes up 10%, then the stock with go up 20%, and if the market declines by 10% then the stock will fall 20%. However, empirical evidence suggests otherwise.

Part of the story is that people forget that the mathematical definition of beta is a combination of the stock's correlation with the market and it's co-variance, or co-variability, with the market. Thus volatile stocks with little market correlation (think health stocks for example) can appear to have modest betas versus stocks that are perfectly correlated and have lower volatility.

I parsed the S&P500 over the last 35 years by beta, dividing the index into deciles, highest beta to lowest. The highest decile averaged a beta of about 1.5, the lowest 0.6. This was the standard 60 month beta calculation, or slope of the regression line through the data plot of the stock's monthly return relative to the market's. Then I looked at how the annualized performance of the extreme deciles compared, and the results tell an interesting story.

Overall, holding an equal weighted portfolio of the 50 highest beta stocks, rebalanced annually, gave an annualized return of about 14%, versus an equal weighted average return of the S&P500 of about 15%, so a slight lag. Yet this underperformance came with tremendous volatility, as the annualized standard deviation of the high beta portfolio was over 28%, versus about 19% for the market. These results compare to the low beta portfolio which averaged over 15% returns with only 15% deviation. So the high beta stocks didn't give a 50% pop (1.5 beta average) in returns, but did so in risk (50% more volatility), so no extra return for the extra risk. The low beta stocks basically matched the market with less risk, a "tie/win" outcome in return/risk lingo. But wait, the results are more interesting depending on market direction.

I looked at how the 2 extreme beta portfolios performed in up and down markets. If 12 month performance of the market was up how much more did the "leveraged" high beta portfolio perform? Well, they did beat the market by a little over 2% per year, while the low beta portfolio lagged by a little less than 2%, yet the results weren't consistent enough to be statistically significant. So you can't definitively expect a high beta portfolio to outperform in a bull market, nor a low beta portfolio to lag. So much for the bull market bet. But what if the market declines?

The average return of of the high beta portfolio in declining markets fell over 13% more than the market, the low beta portfolio outperformed the declining market by about 9%, with both results statistically significant. So high beta stocks crater in falling markets and lower risk, low beta stocks hang tough.

Perhaps none of this is a surprise, but the bottom line is that using high beta stocks for a market bet is a loser's bet - no guarantee of more upside with a rally and near certainty of a bigger downside in a selloff. Given the current tenuous market environment, I suggest that you stay away from the riskiest, high beta stocks, those with a beta greater than 2.0. On this criteria alone, I'd be wary of stocks such as NVIDIA (NVDA, 3.86 beta), Amazon (AMZN, 2.94), Goodyear Tire (GT, 2.06) and Kodak (EK, 2.07). Nothing against these stocks in general, just that their past correlation and co-variance behavior suggest that they won't offer good returns for the extra risk.

Disclosure: The author holds no positions in the above-mentioned stocks.

Comments

[jimmy46]

"definition of beta is a combination of the stock's correlation with the market "

Given that beta is the result of multiplying 2 different numbers,
I've never understood what value it had.


How can you know which of the 2 numbers is more significant when both are hidden in the final result?

It must have been invented by a college prof. who had no money in the market.

[bb99]

Not only does Amazon have a high beta, but is now starting to have headwinds such as states such as New York forcing Amazon to collect taxes, a move that will obviously be copied by all cash-strapped states. Amazon is also losing market share in music and will continue with a tax advantage disappearing. The risk is incrementally greater today!

[T Muller]

The beta coefficient is not related to correlation. A line can be drawn through any set of data points no matter how unrelated the data points. The beta would just be the slope of the linear regression trend line, which is the "best fit" yet it could be a terrible fit relatively.

The R2- or coefficient of determination - explains the amount of variation of Y shared by X. Thus, if r2= .5 then it's explained as 50% of the the variation of StockABC is attributed to the variation of sp500.

The author mentions that some stocks (health care) have modest betas but are very volatile and have a low correlation to the market. This would suggest that the r2 would be 0 to .2 or so. Stocks like GE - which is very similar to the overall market might have .5-.6 or higher R2.

The Beta, or slope, just explains the relationship of Y to X - magnitude of change. Betas of .5 and 5 can be perfectly correlated with the market (r2=1), so beta doesn't explain the goodness of fit, just how x impacts y- If a stock moved 5 times as much than sp500, for every change, then the r2= 1 and the Beta would be 5.

My point is - (after longwinded diatribe) is that some Betas can be worthless. Absolutely worthless- This largely depends on how much the market's move is responsible for moves in the stock, the r2. Thus, that is an important factor to look at to evaluate Beta reasonableness.

There are many different estimations out there for any given stock.

AMZN-
Yahoo-3.3
Google-2.7
Value Line 1.2
MSN-2.95

Depends on the time frame used for the measurement.

[mkreisel]

For high beta stocks, you buy low and sell high (or until the trend is broken). When NVDA dropped to 3 something in 2004, at least 9 insiders swooped in and bought tons of stocks. How much money would you have made had you been paying a little more attention?

Why must everyone buy and hold(hope)?