Trillade över denna artikel av Larry Swedroe som jag gillar:
som har skrivit en artikel på följande studie:
Their sample covered the period from March 31, 1951, through Aug. 31, 2022. Here is a summary of their key findings:
Across the five markets, drawdowns of up to 5% occurred in 38.3% of the months, with 8.2% of the drawdowns occurring during recessions. Drawdowns of 5% to 10% occurred in 18.3% of the months, with 13.4% of the drawdowns occurring during recessions. Drawdowns of 10% to 20% occurred in 19.3% of the months, with 21.8% of the drawdowns occurring during recessions. And drawdowns of more than 20% occurred in 11.9% of the months, with 38.9% of the drawdowns occurring during recessions. The pattern makes clear that while the more-severe drawdowns are more likely to occur during recessions, most do not—the likely explanation is that the market is a leading economic indicator, tending to fall in anticipation of recessions.
Large equity market drawdowns were generally more explainable than small drawdowns. However, as the drawdown size increased, the number of factors that met the threshold of statistical significance increased, as did the R-squared value—it is difficult to predict small equity market declines relative to their larger counterparts.
Since 1951 global markets had returned, on average, 7.6% over cash and experienced an average drawdown of 15.0% over a one-year horizon. However, on average, when the drawdown probability was above 30%, global markets performed poorly over the following year, with an excess return of negative 1.7% and an average drawdown of 22%. For probabilities below 30%, the average excess market return and drawdown were 11.4% and 12.2%, respectively.
Market crashes historically have been associated with a set of factors centered around valuation (particularly dividend yield), technical (the 12-month Sharpe ratio), and macroeconomic (inflation and credit growth) indicators.
Their model showed a likelihood of a crash of more than 80% prior to the crashes in 1973 and 2000, and more than 60% prior to the crash that began in late 2007. In 2021 the likelihood of a crash was less than 20% but by August 2022 had risen to above 80%.
Their framework was effective for both recessionary and nonrecessionary drawdowns.
Valuation was the only factor that showed a high degree of statistical significance across all markets except Japan (for which no factor was statistically significant). However, by pooling data (and excluding Japan) almost all factors were statistically significant, with inflation being the most significant (t-stat = 5.34), followed by valuation (t-stat = negative 4.88).
While the U.S. was a fundamental driver of “market fragility” globally—the incremental effect of a U.S. drawdown was 0.31, implying a 31% increase in each country’s conditional drawdown probability if the U.S. equity market were impaired—country-specific factors were still relevant for predicting the likelihood of large equity market drawdowns.
The finding that valuations play an important role in forecasting future equity returns and the risk of large market drawdowns is consistent with prior literature, including the 2012 study “An Old Friend: The Stock Market’s Shiller P/E,” the 2022 study “Equity Risk Premiums (ERP): Determinants, Estimation, and Implications,” and the 2023 study “The Continued Forecasting Effectiveness of a Real Earnings Model of the Equity Premium.”
Ping @zino, @Andre_Granstrom