Are stocks really riskier than bonds?

Abstract

Conventional wisdom holds that stocks are riskier than bonds; thus when the stock market becomes volatile, money flows from the stock market into the perceived safe haven of the bond market. In this article, we find that this notion is not necessarily accurate and might lead people to make incorrect investment decisions. In fact, intermediate- and long-term bonds are riskier than stocks when we measure risk by the coefficient of variation. We examine a case where an inaccurate perception regarding the relative riskiness of the two types of assets could play a part in what appears to be short-sighted and potentially costly behaviour of investors in financial markets.

Notes

¹ See Page 282.

² See Page 387.

³ See Page 156.

⁴ See Page 144.

⁵ See Page 294.

⁶ See Page 686.

⁷ See, for example, Mayo (2003).

⁸ Typically, 10-year Treasuries are known as intermediate-term bonds and 20-year T-bonds are long-term bonds.

⁹ The data for T-bonds and T-bills are from CRSP. Monthly total return data from April 1953 to September 2003 are used since this period overlaps the S&P 500 total return data via Standard & Poors Corp. We do not include corporate bonds in comparison since the equivalent data (total return data) are not available for corporate bonds.

¹⁰ Breunig (2001) demonstrates that CV estimates are biased in skewed distributions. In negatively skewed distributions, like that of S&P returns, the CV is overestimated, and in positively skewed returns such as bonds, the bias is the opposite. This indicates that the relative risk of bonds to stocks may be even greater than implied by our CV estimates. See Table 1 for return distribution information and bias-adjusted CV estimates.

¹¹ Most mutual funds fall into one of three categories: stock funds, which invest in equity securities, bond funds, which invest in fixed income securities, and money market funds, which invest in very short-term interest-paying assets. If investors do not understand relative risk and reward or basic interest rate and pricing relationships, their strategies in timing purchases and sales as well as in asset allocation may result in sub-optimal returns on mutual fund assets and could even create problems for fund managers who are forced to take or reduce positions at the wrong time.

¹² We use the Aaa interest rate instead of Treasuries rates in this section since the majority of bond funds under consideration are composed of corporate bonds.

¹³ Previous literature suggests that returns on common stock (e.g., Chen et al. (1986)) and bonds (e.g., Elton et al. (1995)) can be explained by variables such as market returns, risk, term premium, inflation and economic performance.

¹⁴ The original autoregressive conditional heteroskedasticity (ARCH) model was developed by Engle (1982). Bollerslev (1986) extends the ARCH model to a GARCH model. The GARCH model differs from the ARCH model in that the former conditions, current variance on lagged variance, as is shown in the specification of our model.

¹⁵ These data can be found at http://research.stlouisfed.org/fred2/.