***Stock market declines can collapse the economy

(Robert J. Barro – PhD in Economics @ Harvard University, Paul M. Warburg Professor of Economics @ Harvard University, senior fellow of the Hoover Institution of Stanford University, and a research associate of the National Bureau of Economic Research. José F.Ursúa – PhD candidate in Economics @ Harvard University, “Stock-market crashes and depressions,” Research In Economics, Volume 71, Issue 3, September 2017, ScienceDirect)

5. Conclusions and future research The long-term history for 30 countries indicates that stock-market crashes (cumulative real returns of –0.25 or worse) have substantial predictive power for depressions (cumulative macroeconomic declines by 10 percent or more). In a non-war environment, the realization of a stock-market return of –0.25 or worse implies that the probability of a minor depression (macroeconomic decline by 10 percent or more) is 22 percent, and the probability of a major depression (decline by 25 percent or more) is 3 percent. When the stock return is –0.40 or worse, the corresponding depression probabilities are 30 percent and 4 percent. The probabilities are 46 percent and 8 percent, respectively, when a stock-market crash of –0.25 or worse in a non-war period is accompanied by a currency or banking crisis and occurs during a time of global economic turmoil. Given the stock-market crashes for the United States and other countries in 2008 (characterized as non-war, featuring a currency or banking crisis, and occurring during a global event), these last probabilities applied at the beginning of 2009. The long-term data also show that the majority (67 percent) of minor, non-war depressions are accompanied by stock-market crashes, whereas most major, non-war depressions (83 percent) are accompanied by these crashes. Therefore, in the absence of a stock-market crash, the occurrence of a depression is highly unlikely. The full history of 3037 annual observations reveals 71 matched cases of stock-market crashes and depressions, of which 41 percent are associated with war. The years contained in the 71 cases constitute 12 percent of the overall sample (because the average duration of a crisis was 4.9 years). The co-movement between stock returns and macroeconomic changes for the 71 cases may be sufficient to explain the observed equity premium of 7 percent, assuming that the coefficient of relative risk aversion is between 3.5 and 4. The required coefficient falls to 3.5 or less if we bring in the additional 19 percent of the sample that features a stock-market crash or depression but not both. A crucial aspect of this analysis is that the computed covariances allow for a flexible timing pattern between stock returns and macroeconomic changes. In future research, we plan to assess this flexible covariance approach in more detail. Part of this research involves the roles of missing data and wartime price controls, which can distort the covariances calculated from a rigid timing structure.