Goldman Sachs has been producing a lot of bullish equity research, despite chief U.S. equity strategist David Kostin’s bearish view on stocks. Last month, Goldman’s Peter Oppenheimer published a massive report titled The Long Good Buy that argued stocks were insanely cheap relative to bonds.
In a new report today title Houses vs. Stocks: Who Wins the Long-Run ‘Sharpe’ Race?, Goldman’s Jose Ursua argues that stocks are way better than housing on both an absolute returns basis and risk-adjusted returns basis.
One of the most common risk-adjusted returns measures is the Sharpe ratio. Simply put, it’s the excess returns divided by the standard deviation of excess returns. For long-term investments like housing or stocks, Ursua argues “What matters are risk-reward tradeoffs.”
Here’s a summary of his report.
Goldman’s research considers stock market data and home price data (for Australia, Canada, Finland, France, Iceland, Japan, Netherlands, Norway, Spain, UK, and US) going back to 1890.
Since 1890, the global average of yearly housing returns is approximately 1.7%, while for stocks it is 5.6%. The difference is smaller for the relatively more tumultuous pre-1948 period (1.5% vs. 4.0%, respectively) and larger during the post-1948 period (at 1.7% vs. 7.0%, respectively). On average, the magnitude of stock returns is larger than for house returns by a factor of 2-4.
But according to that same data, stocks are much more volatile.
Again since 1890, the standard deviation of the global average of yearly housing returns is 5.5%, while that of stocks is 15%. Housing returns were particularly volatile during the pre-1948 period (at 7.3% vs. 3.2% after that), while the volatility of stocks actually increased post-WWII (from 13% to 17%). Generally speaking then, stocks are more volatile than housing by a factor of 2-5.
Sharpe Ratio (Risk-Adjusted Return)
Without getting into too many details, Sharpe ratio is higher for stocks than housing. Here’s Ursua’s summary:
An empirical way to assess the Sharpe Ratio is to approximate the actual measure by using returns on short-term bills as a proxy for the risk-free rate. This allows computing both the numerator (mean excess returns) and the denominator (standard deviation of excess returns). The mean and median real returns on bills are at around 1%, with a standard deviation of around 9%, although the mean is even negative in some sub-samples. Putting all this together, for housing returns, we get an estimated Sharpe Ratio of 0.16 for the global portfolio (0.21 for the pre-1948 period, and 0.12 afterward). For stock returns, that figure is 0.37 (0.39 for the pre-1948 period, and 0.36 afterward). Therefore, the empirical estimates of the Sharpe Ratio are higher than the theoretical ones and show a better reward-to-variability tradeoff for stocks than for housing.
Ursua warns that there are many other variables that should be consider when calculating housing market returns (e.g. leverage, rental yields, and taxes). He also notes that there are other measures of risk-reward tradeoffs other than the Sharpe ratio.
But based on this analysis stocks are the better buy.